external/ton
1
0
Fork 0
mirror of https://github.com/ton-blockchain/ton synced 2025-02-14 12:12:21 +00:00
Code Issues Releases Wiki Activity
ton/tolk/ast-from-tokens.cpp

1165 lines
40 KiB
C++
Raw Permalink Normal View History

[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
/*
This file is part of TON Blockchain Library.
TON Blockchain Library is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.
TON Blockchain Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with TON Blockchain Library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "ast-from-tokens.h"
#include "ast.h"
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
#include "type-system.h"
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
#include "platform-utils.h"
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
#include "tolk-version.h"
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
/*
* Here we construct AST for a tolk file.
* While constructing, no global state is modified.
* Historically, in FunC, there was no AST: while lexing, symbols were registered, types were inferred, and so on.
* There was no way to perform any more or less semantic analysis.
* Implementing AST gives a giant advance for future modifications and stability.
*/
namespace tolk {
// given a token, determine whether it's <, or >, or similar
static bool is_comparison_binary_op(TokenType tok) {
return tok == tok_lt || tok == tok_gt || tok == tok_leq || tok == tok_geq || tok == tok_eq || tok == tok_neq || tok == tok_spaceship;
}
// same as above, but to detect bitwise operators: & | ^
static bool is_bitwise_binary_op(TokenType tok) {
return tok == tok_bitwise_and || tok == tok_bitwise_or || tok == tok_bitwise_xor;
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// same as above, but to detect logical operators: && ||
static bool is_logical_binary_op(TokenType tok) {
return tok == tok_logical_and || tok == tok_logical_or;
}
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
// same as above, but to detect addition/subtraction
static bool is_add_or_sub_binary_op(TokenType tok) {
return tok == tok_plus || tok == tok_minus;
}
// fire an error for a case "flags & 0xFF != 0" (equivalent to "flags & 1", probably unexpected)
// it would better be a warning, but we decided to make it a strict error
GNU_ATTRIBUTE_NORETURN GNU_ATTRIBUTE_COLD
static void fire_error_lower_precedence(SrcLocation loc, std::string_view op_lower, std::string_view op_higher) {
std::string name_lower = static_cast<std::string>(op_lower);
std::string name_higher = static_cast<std::string>(op_higher);
throw ParseError(loc, name_lower + " has lower precedence than " + name_higher +
", probably this code won't work as you expected. "
"Use parenthesis: either (... " + name_lower + " ...) to evaluate it first, or (... " + name_higher + " ...) to suppress this error.");
}
// fire an error for a case "arg1 & arg2 | arg3"
GNU_ATTRIBUTE_NORETURN GNU_ATTRIBUTE_COLD
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
static void fire_error_mix_and_or_no_parenthesis(SrcLocation loc, std::string_view op1, std::string_view op2) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
std::string name1 = static_cast<std::string>(op1);
std::string name2 = static_cast<std::string>(op2);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
throw ParseError(loc, "mixing " + name1 + " with " + name2 + " without parenthesis may lead to accidental errors. "
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
"Use parenthesis to emphasize operator precedence.");
}
// diagnose when bitwise operators are used in a probably wrong way due to tricky precedence
// example: "flags & 0xFF != 0" is equivalent to "flags & 1", most likely it's unexpected
// the only way to suppress this error for the programmer is to use parenthesis
// (how do we detect presence of parenthesis? simple: (0!=1) is ast_parenthesized_expr{ast_binary_operator},
// that's why if rhs->type == ast_binary_operator, it's not surrounded by parenthesis)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static void diagnose_bitwise_precedence(SrcLocation loc, std::string_view operator_name, AnyExprV lhs, AnyExprV rhs) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
// handle "flags & 0xFF != 0" (rhs = "0xFF != 0")
if (rhs->type == ast_binary_operator && is_comparison_binary_op(rhs->as<ast_binary_operator>()->tok)) {
fire_error_lower_precedence(loc, operator_name, rhs->as<ast_binary_operator>()->operator_name);
}
// handle "0 != flags & 0xFF" (lhs = "0 != flags")
if (lhs->type == ast_binary_operator && is_comparison_binary_op(lhs->as<ast_binary_operator>()->tok)) {
fire_error_lower_precedence(loc, operator_name, lhs->as<ast_binary_operator>()->operator_name);
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
// similar to above, but detect potentially invalid usage of && and ||
// since anyway, using parenthesis when both && and || occur in the same expression,
// && and || have equal operator precedence in Tolk
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static void diagnose_and_or_precedence(SrcLocation loc, AnyExprV lhs, TokenType rhs_tok, std::string_view rhs_operator_name) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (auto lhs_op = lhs->try_as<ast_binary_operator>()) {
// handle "arg1 & arg2 | arg3" (lhs = "arg1 & arg2")
if (is_bitwise_binary_op(lhs_op->tok) && is_bitwise_binary_op(rhs_tok) && lhs_op->tok != rhs_tok) {
fire_error_mix_and_or_no_parenthesis(loc, lhs_op->operator_name, rhs_operator_name);
}
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// handle "arg1 && arg2 || arg3" (lhs = "arg1 && arg2")
if (is_logical_binary_op(lhs_op->tok) && is_logical_binary_op(rhs_tok) && lhs_op->tok != rhs_tok) {
fire_error_mix_and_or_no_parenthesis(loc, lhs_op->operator_name, rhs_operator_name);
}
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
}
// diagnose "a << 8 + 1" (equivalent to "a << 9", probably unexpected)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static void diagnose_addition_in_bitshift(SrcLocation loc, std::string_view bitshift_operator_name, AnyExprV rhs) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
if (rhs->type == ast_binary_operator && is_add_or_sub_binary_op(rhs->as<ast_binary_operator>()->tok)) {
fire_error_lower_precedence(loc, bitshift_operator_name, rhs->as<ast_binary_operator>()->operator_name);
}
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// replace (a == null) and similar to isNull(a) (call of a built-in function)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV maybe_replace_eq_null_with_isNull_call(V<ast_binary_operator> v) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
bool has_null = v->get_lhs()->type == ast_null_keyword || v->get_rhs()->type == ast_null_keyword;
bool replace = has_null && (v->tok == tok_eq || v->tok == tok_neq);
if (!replace) {
return v;
}
auto v_ident = createV<ast_identifier>(v->loc, "__isNull"); // built-in function
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
auto v_ref = createV<ast_reference>(v->loc, v_ident, nullptr);
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV v_null = v->get_lhs()->type == ast_null_keyword ? v->get_rhs() : v->get_lhs();
AnyExprV v_arg = createV<ast_argument>(v->loc, v_null, false);
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
AnyExprV v_isNull = createV<ast_function_call>(v->loc, v_ref, createV<ast_argument_list>(v->loc, {v_arg}));
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (v->tok == tok_neq) {
v_isNull = createV<ast_unary_operator>(v->loc, "!", tok_logical_not, v_isNull);
}
return v_isNull;
}
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
/*
*
* PARSE SOURCE
*
*/
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV parse_expr(Lexer& lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
static AnyV parse_parameter(Lexer& lex, bool is_first) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
// optional keyword `mutate` meaning that a function will mutate a passed argument (like passed by reference)
bool declared_as_mutate = false;
if (lex.tok() == tok_mutate) {
lex.next();
declared_as_mutate = true;
}
// parameter name (or underscore for an unnamed parameter)
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
std::string_view param_name;
if (lex.tok() == tok_identifier) {
param_name = lex.cur_str();
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
} else if (lex.tok() == tok_self) {
if (!is_first) {
lex.error("`self` can only be the first parameter");
}
param_name = "self";
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
} else if (lex.tok() != tok_underscore) {
lex.unexpected("parameter name");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
lex.next();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
// parameter type after colon are mandatory
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_colon, "`: <parameter_type>`");
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
TypePtr param_type = parse_type_from_tokens(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
return createV<ast_parameter>(loc, param_name, param_type, declared_as_mutate);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
static AnyV parse_global_var_declaration(Lexer& lex, const std::vector<V<ast_annotation>>& annotations) {
if (!annotations.empty()) {
lex.error("@annotations are not applicable to global var declaration");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
SrcLocation loc = lex.cur_location();
lex.expect(tok_global, "`global`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.check(tok_identifier, "global variable name");
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
auto v_ident = createV<ast_identifier>(lex.cur_location(), lex.cur_str());
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_colon, "`:`");
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
TypePtr declared_type = parse_type_from_tokens(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_comma) {
lex.error("multiple declarations are not allowed, split globals on separate lines");
}
if (lex.tok() == tok_assign) {
lex.error("assigning to a global is not allowed at declaration");
}
lex.expect(tok_semicolon, "`;`");
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
return createV<ast_global_var_declaration>(loc, v_ident, declared_type);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
static AnyV parse_constant_declaration(Lexer& lex, const std::vector<V<ast_annotation>>& annotations) {
if (!annotations.empty()) {
lex.error("@annotations are not applicable to global var declaration");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
SrcLocation loc = lex.cur_location();
lex.expect(tok_const, "`const`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.check(tok_identifier, "constant name");
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
auto v_ident = createV<ast_identifier>(lex.cur_location(), lex.cur_str());
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
TypePtr declared_type = nullptr;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_colon) {
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
declared_type = parse_type_from_tokens(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
lex.expect(tok_assign, "`=`");
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV init_value = parse_expr(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_comma) {
lex.error("multiple declarations are not allowed, split constants on separate lines");
}
lex.expect(tok_semicolon, "`;`");
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
return createV<ast_constant_declaration>(loc, v_ident, declared_type, init_value);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
// "parameters" are at function declaration: `fun f(param1: int, mutate param2: slice)`
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
static V<ast_parameter_list> parse_parameter_list(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
std::vector<AnyV> params;
lex.expect(tok_oppar, "parameter list");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
if (lex.tok() != tok_clpar) {
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
params.push_back(parse_parameter(lex, true));
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
while (lex.tok() == tok_comma) {
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
params.push_back(parse_parameter(lex, false));
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_clpar, "`)`");
return createV<ast_parameter_list>(loc, std::move(params));
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
// "arguments" are at function call: `f(arg1, mutate arg2)`
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_argument(Lexer& lex) {
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
SrcLocation loc = lex.cur_location();
// keyword `mutate` is necessary when a parameter is declared `mutate` (to make mutation obvious for the reader)
bool passed_as_mutate = false;
if (lex.tok() == tok_mutate) {
lex.next();
passed_as_mutate = true;
}
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV expr = parse_expr(lex);
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
return createV<ast_argument>(loc, expr, passed_as_mutate);
}
static V<ast_argument_list> parse_argument_list(Lexer& lex) {
SrcLocation loc = lex.cur_location();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
std::vector<AnyExprV> args;
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
lex.expect(tok_oppar, "`(`");
if (lex.tok() != tok_clpar) {
args.push_back(parse_argument(lex));
while (lex.tok() == tok_comma) {
lex.next();
args.push_back(parse_argument(lex));
}
}
lex.expect(tok_clpar, "`)`");
return createV<ast_argument_list>(loc, std::move(args));
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
static V<ast_instantiationT_list> parse_maybe_instantiationTs_after_identifier(Lexer& lex) {
lex.check(tok_lt, "`<`");
Lexer::SavedPositionForLookahead backup = lex.save_parsing_position();
try {
SrcLocation loc = lex.cur_location();
lex.next();
std::vector<AnyV> instantiationTs;
instantiationTs.push_back(createV<ast_instantiationT_item>(lex.cur_location(), parse_type_from_tokens(lex)));
while (lex.tok() == tok_comma) {
lex.next();
instantiationTs.push_back(createV<ast_instantiationT_item>(lex.cur_location(), parse_type_from_tokens(lex)));
}
lex.expect(tok_gt, "`>`");
return createV<ast_instantiationT_list>(loc, std::move(instantiationTs));
} catch (const ParseError&) {
lex.restore_position(backup);
return nullptr;
}
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// parse (expr) / [expr] / identifier / number
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr100(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
switch (lex.tok()) {
case tok_oppar: {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_clpar) {
lex.next();
return createV<ast_tensor>(loc, {});
}
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV first = parse_expr(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_clpar) {
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
return createV<ast_parenthesized_expression>(loc, first);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
std::vector<AnyExprV> items(1, first);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
while (lex.tok() == tok_comma) {
lex.next();
items.emplace_back(parse_expr(lex));
}
lex.expect(tok_clpar, "`)`");
return createV<ast_tensor>(loc, std::move(items));
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
case tok_opbracket: {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_clbracket) {
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
return createV<ast_typed_tuple>(loc, {});
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
std::vector<AnyExprV> items(1, parse_expr(lex));
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
while (lex.tok() == tok_comma) {
lex.next();
items.emplace_back(parse_expr(lex));
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_clbracket, "`]`");
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
return createV<ast_typed_tuple>(loc, std::move(items));
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
case tok_int_const: {
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
std::string_view orig_str = lex.cur_str();
td::RefInt256 intval = td::string_to_int256(static_cast<std::string>(orig_str));
if (intval.is_null() || !intval->signed_fits_bits(257)) {
lex.error("invalid integer constant");
}
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
return createV<ast_int_const>(loc, std::move(intval), orig_str);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
case tok_string_const: {
std::string_view str_val = lex.cur_str();
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
char modifier = 0;
if (lex.tok() == tok_string_modifier) {
modifier = lex.cur_str()[0];
lex.next();
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
return createV<ast_string_const>(loc, str_val, modifier);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
case tok_underscore: {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
return createV<ast_underscore>(loc);
}
case tok_true: {
lex.next();
return createV<ast_bool_const>(loc, true);
}
case tok_false: {
lex.next();
return createV<ast_bool_const>(loc, false);
}
case tok_null: {
lex.next();
return createV<ast_null_keyword>(loc);
}
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
case tok_self: {
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
auto v_ident = createV<ast_identifier>(loc, "self");
return createV<ast_reference>(loc, v_ident, nullptr);
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
case tok_identifier: {
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
auto v_ident = createV<ast_identifier>(loc, lex.cur_str());
V<ast_instantiationT_list> v_instantiationTs = nullptr;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
if (lex.tok() == tok_lt) {
v_instantiationTs = parse_maybe_instantiationTs_after_identifier(lex);
}
return createV<ast_reference>(loc, v_ident, v_instantiationTs);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
[Tolk] Allow `cell` and `slice` be valid identifiers They are not keywords anymore. > var cell = ...; > var cell: cell = ...; Motivation: in the future, when structures are implemented, this obviously should be valid: > struct a { ... } > var a = ...; Struct fields will also be allowed to have names int/slice/cell.
2025-01-27 07:33:24 +00:00
default:
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.unexpected("<expression>");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
// parse E(...) (left-to-right)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr90(Lexer& lex) {
AnyExprV res = parse_expr100(lex);
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
while (lex.tok() == tok_oppar) {
res = createV<ast_function_call>(res->loc, res, parse_argument_list(lex));
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
return res;
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
// parse E.field and E.method(...) (left-to-right)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr80(Lexer& lex) {
AnyExprV lhs = parse_expr90(lex);
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
while (lex.tok() == tok_dot) {
SrcLocation loc = lex.cur_location();
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
V<ast_identifier> v_ident = nullptr;
V<ast_instantiationT_list> v_instantiationTs = nullptr;
[Tolk] Support syntax `tensorVar.0` and `tupleVar.0` It works both for reading and writing: > var t = (1, 2); > t.0; // 1 > t.0 = 5; > t; // (5, 2) It also works for typed/untyped tuples, producing INDEX and SETINDEX. Global tensors and tuples works. Nesting `t.0.1.2` works. `mutate` works. Even mixing tuples inside tensors inside a global for writing works.
2025-01-27 07:29:17 +00:00
if (lex.tok() == tok_identifier) { // obj.field / obj.method
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
v_ident = createV<ast_identifier>(lex.cur_location(), lex.cur_str());
lex.next();
if (lex.tok() == tok_lt) {
v_instantiationTs = parse_maybe_instantiationTs_after_identifier(lex);
}
[Tolk] Support syntax `tensorVar.0` and `tupleVar.0` It works both for reading and writing: > var t = (1, 2); > t.0; // 1 > t.0 = 5; > t; // (5, 2) It also works for typed/untyped tuples, producing INDEX and SETINDEX. Global tensors and tuples works. Nesting `t.0.1.2` works. `mutate` works. Even mixing tuples inside tensors inside a global for writing works.
2025-01-27 07:29:17 +00:00
} else if (lex.tok() == tok_int_const) { // obj.0 (indexed access)
v_ident = createV<ast_identifier>(lex.cur_location(), lex.cur_str());
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
} else {
lex.unexpected("method name");
}
lhs = createV<ast_dot_access>(loc, lhs, v_ident, v_instantiationTs);
while (lex.tok() == tok_oppar) {
lhs = createV<ast_function_call>(lex.cur_location(), lhs, parse_argument_list(lex));
}
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
return lhs;
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// parse ! ~ - + E (unary)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr75(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
TokenType t = lex.tok();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (t == tok_logical_not || t == tok_bitwise_not || t == tok_minus || t == tok_plus) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
std::string_view operator_name = lex.cur_str();
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV rhs = parse_expr75(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
return createV<ast_unary_operator>(loc, operator_name, t, rhs);
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
return parse_expr80(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
// parse E as <type>
static AnyExprV parse_expr40(Lexer& lex) {
AnyExprV lhs = parse_expr75(lex);
if (lex.tok() == tok_as) {
SrcLocation loc = lex.cur_location();
lex.next();
TypePtr cast_to_type = parse_type_from_tokens(lex);
lhs = createV<ast_cast_as_operator>(loc, lhs, cast_to_type);
}
return lhs;
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// parse E * / % ^/ ~/ E (left-to-right)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr30(Lexer& lex) {
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
AnyExprV lhs = parse_expr40(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
TokenType t = lex.tok();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
while (t == tok_mul || t == tok_div || t == tok_mod || t == tok_divC || t == tok_divR) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
std::string_view operator_name = lex.cur_str();
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
AnyExprV rhs = parse_expr40(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lhs = createV<ast_binary_operator>(loc, operator_name, t, lhs, rhs);
t = lex.tok();
}
return lhs;
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// parse E + - E (left-to-right)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr20(Lexer& lex) {
AnyExprV lhs = parse_expr30(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
TokenType t = lex.tok();
while (t == tok_minus || t == tok_plus) {
SrcLocation loc = lex.cur_location();
std::string_view operator_name = lex.cur_str();
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV rhs = parse_expr30(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lhs = createV<ast_binary_operator>(loc, operator_name, t, lhs, rhs);
t = lex.tok();
}
return lhs;
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// parse E << >> ~>> ^>> E (left-to-right)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr17(Lexer& lex) {
AnyExprV lhs = parse_expr20(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
TokenType t = lex.tok();
while (t == tok_lshift || t == tok_rshift || t == tok_rshiftC || t == tok_rshiftR) {
SrcLocation loc = lex.cur_location();
std::string_view operator_name = lex.cur_str();
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV rhs = parse_expr20(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
diagnose_addition_in_bitshift(loc, operator_name, rhs);
lhs = createV<ast_binary_operator>(loc, operator_name, t, lhs, rhs);
t = lex.tok();
}
return lhs;
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// parse E == < > <= >= != <=> E (left-to-right)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr15(Lexer& lex) {
AnyExprV lhs = parse_expr17(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
TokenType t = lex.tok();
if (t == tok_eq || t == tok_lt || t == tok_gt || t == tok_leq || t == tok_geq || t == tok_neq || t == tok_spaceship) {
SrcLocation loc = lex.cur_location();
std::string_view operator_name = lex.cur_str();
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV rhs = parse_expr17(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lhs = createV<ast_binary_operator>(loc, operator_name, t, lhs, rhs);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (t == tok_eq || t == tok_neq) {
lhs = maybe_replace_eq_null_with_isNull_call(lhs->as<ast_binary_operator>());
}
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
return lhs;
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// parse E & | ^ E (left-to-right)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr14(Lexer& lex) {
AnyExprV lhs = parse_expr15(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
TokenType t = lex.tok();
while (t == tok_bitwise_and || t == tok_bitwise_or || t == tok_bitwise_xor) {
SrcLocation loc = lex.cur_location();
std::string_view operator_name = lex.cur_str();
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV rhs = parse_expr15(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
diagnose_bitwise_precedence(loc, operator_name, lhs, rhs);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
diagnose_and_or_precedence(loc, lhs, t, operator_name);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lhs = createV<ast_binary_operator>(loc, operator_name, t, lhs, rhs);
t = lex.tok();
}
return lhs;
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// parse E && || E (left-to-right)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr13(Lexer& lex) {
AnyExprV lhs = parse_expr14(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
TokenType t = lex.tok();
while (t == tok_logical_and || t == tok_logical_or) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
std::string_view operator_name = lex.cur_str();
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV rhs = parse_expr14(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
diagnose_and_or_precedence(loc, lhs, t, operator_name);
lhs = createV<ast_binary_operator>(loc, operator_name, t, lhs, rhs);
t = lex.tok();
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
return lhs;
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// parse E = += -= E and E ? E : E (right-to-left)
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_expr10(Lexer& lex) {
AnyExprV lhs = parse_expr13(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
TokenType t = lex.tok();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
if (t == tok_assign) {
SrcLocation loc = lex.cur_location();
lex.next();
AnyExprV rhs = parse_expr10(lex);
return createV<ast_assign>(loc, lhs, rhs);
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (t == tok_set_plus || t == tok_set_minus || t == tok_set_mul || t == tok_set_div ||
t == tok_set_mod || t == tok_set_lshift || t == tok_set_rshift ||
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
t == tok_set_bitwise_and || t == tok_set_bitwise_or || t == tok_set_bitwise_xor) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
std::string_view operator_name = lex.cur_str().substr(0, lex.cur_str().size() - 1); // "+" for +=
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV rhs = parse_expr10(lex);
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
return createV<ast_set_assign>(loc, operator_name, t, lhs, rhs);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (t == tok_question) {
SrcLocation loc = lex.cur_location();
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV when_true = parse_expr10(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_colon, "`:`");
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV when_false = parse_expr10(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
return createV<ast_ternary_operator>(loc, lhs, when_true, when_false);
}
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
return lhs;
}
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV parse_expr(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
return parse_expr10(lex);
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
AnyV parse_statement(Lexer& lex);
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_var_declaration_lhs(Lexer& lex, bool is_immutable) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_oppar) {
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV first = parse_var_declaration_lhs(lex, is_immutable);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_clpar) {
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
return first;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
std::vector<AnyExprV> args(1, first);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
while (lex.tok() == tok_comma) {
lex.next();
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
args.push_back(parse_var_declaration_lhs(lex, is_immutable));
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
lex.expect(tok_clpar, "`)`");
return createV<ast_tensor>(loc, std::move(args));
}
if (lex.tok() == tok_opbracket) {
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
std::vector<AnyExprV> args(1, parse_var_declaration_lhs(lex, is_immutable));
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
while (lex.tok() == tok_comma) {
lex.next();
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
args.push_back(parse_var_declaration_lhs(lex, is_immutable));
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
lex.expect(tok_clbracket, "`]`");
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
return createV<ast_typed_tuple>(loc, std::move(args));
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
if (lex.tok() == tok_identifier) {
auto v_ident = createV<ast_identifier>(loc, lex.cur_str());
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
TypePtr declared_type = nullptr;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
bool marked_as_redef = false;
lex.next();
if (lex.tok() == tok_colon) {
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
declared_type = parse_type_from_tokens(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
} else if (lex.tok() == tok_redef) {
lex.next();
marked_as_redef = true;
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
return createV<ast_local_var_lhs>(loc, v_ident, declared_type, is_immutable, marked_as_redef);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
if (lex.tok() == tok_underscore) {
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
TypePtr declared_type = nullptr;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.next();
if (lex.tok() == tok_colon) {
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
declared_type = parse_type_from_tokens(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
return createV<ast_local_var_lhs>(loc, createV<ast_identifier>(loc, ""), declared_type, true, false);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
lex.unexpected("variable name");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
static AnyV parse_local_vars_declaration_assignment(Lexer& lex) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
bool is_immutable = lex.tok() == tok_val;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
AnyExprV lhs = createV<ast_local_vars_declaration>(loc, parse_var_declaration_lhs(lex, is_immutable));
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() != tok_assign) {
lex.error("variables declaration must be followed by assignment: `var xxx = ...`");
}
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
AnyExprV rhs = parse_expr(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_comma) {
lex.error("multiple declarations are not allowed, split variables on separate lines");
}
lex.expect(tok_semicolon, "`;`");
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
return createV<ast_assign>(loc, lhs, rhs);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
static V<ast_sequence> parse_sequence(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_opbrace, "`{`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
std::vector<AnyV> items;
while (lex.tok() != tok_clbrace) {
items.push_back(parse_statement(lex));
}
SrcLocation loc_end = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_clbrace, "`}`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
return createV<ast_sequence>(loc, loc_end, items);
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
static AnyV parse_return_statement(Lexer& lex) {
SrcLocation loc = lex.cur_location();
lex.expect(tok_return, "`return`");
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
AnyExprV child = lex.tok() == tok_semicolon // `return;` actually means "nothing" (inferred as void)
? createV<ast_empty_expression>(lex.cur_location())
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
: parse_expr(lex);
lex.expect(tok_semicolon, "`;`");
return createV<ast_return_statement>(loc, child);
}
[Tolk] `bool` type (-1/0 int under the hood) Comparison operators `== / >= /...` return `bool`. Logical operators `&& ||` return bool. Constants `true` and `false` have the `bool` type. Lots of stdlib functions return `bool`, not `int`. Operator `!x` supports both `int` and `bool`. Condition of `if` accepts both `int` and `bool`. Arithmetic operators are restricted to integers. Logical `&&` and `||` accept both `bool` and `int`. No arithmetic operations with bools allowed (only bitwise and logical).
2025-01-13 08:21:24 +00:00
static AnyV parse_if_statement(Lexer& lex) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
SrcLocation loc = lex.cur_location();
lex.expect(tok_if, "`if`");
lex.expect(tok_oppar, "`(`");
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV cond = parse_expr(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_clpar, "`)`");
V<ast_sequence> if_body = parse_sequence(lex);
V<ast_sequence> else_body = nullptr;
if (lex.tok() == tok_else) { // else if(e) { } or else { }
lex.next();
if (lex.tok() == tok_if) {
[Tolk] `bool` type (-1/0 int under the hood) Comparison operators `== / >= /...` return `bool`. Logical operators `&& ||` return bool. Constants `true` and `false` have the `bool` type. Lots of stdlib functions return `bool`, not `int`. Operator `!x` supports both `int` and `bool`. Condition of `if` accepts both `int` and `bool`. Arithmetic operators are restricted to integers. Logical `&&` and `||` accept both `bool` and `int`. No arithmetic operations with bools allowed (only bitwise and logical).
2025-01-13 08:21:24 +00:00
AnyV v_inner_if = parse_if_statement(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
else_body = createV<ast_sequence>(v_inner_if->loc, lex.cur_location(), {v_inner_if});
} else {
else_body = parse_sequence(lex);
}
} else { // no 'else', create empty block
else_body = createV<ast_sequence>(lex.cur_location(), lex.cur_location(), {});
}
[Tolk] `bool` type (-1/0 int under the hood) Comparison operators `== / >= /...` return `bool`. Logical operators `&& ||` return bool. Constants `true` and `false` have the `bool` type. Lots of stdlib functions return `bool`, not `int`. Operator `!x` supports both `int` and `bool`. Condition of `if` accepts both `int` and `bool`. Arithmetic operators are restricted to integers. Logical `&&` and `||` accept both `bool` and `int`. No arithmetic operations with bools allowed (only bitwise and logical).
2025-01-13 08:21:24 +00:00
return createV<ast_if_statement>(loc, false, cond, if_body, else_body);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
static AnyV parse_repeat_statement(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_repeat, "`repeat`");
lex.expect(tok_oppar, "`(`");
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV cond = parse_expr(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_clpar, "`)`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
V<ast_sequence> body = parse_sequence(lex);
return createV<ast_repeat_statement>(loc, cond, body);
}
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
static AnyV parse_while_statement(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_while, "`while`");
lex.expect(tok_oppar, "`(`");
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV cond = parse_expr(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_clpar, "`)`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
V<ast_sequence> body = parse_sequence(lex);
return createV<ast_while_statement>(loc, cond, body);
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
static AnyV parse_do_while_statement(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_do, "`do`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
V<ast_sequence> body = parse_sequence(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_while, "`while`");
lex.expect(tok_oppar, "`(`");
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV cond = parse_expr(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_clpar, "`)`");
lex.expect(tok_semicolon, "`;`");
return createV<ast_do_while_statement>(loc, body, cond);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
static AnyExprV parse_catch_variable(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_identifier) {
std::string_view var_name = lex.cur_str();
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
auto v_ident = createV<ast_identifier>(loc, var_name);
return createV<ast_reference>(loc, v_ident, nullptr);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
if (lex.tok() == tok_underscore) {
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
auto v_ident = createV<ast_identifier>(loc, "");
return createV<ast_reference>(loc, v_ident, nullptr);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
lex.unexpected("identifier");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
static AnyExprV create_catch_underscore_variable(const Lexer& lex) {
auto v_ident = createV<ast_identifier>(lex.cur_location(), "");
return createV<ast_reference>(lex.cur_location(), v_ident, nullptr);
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
static AnyV parse_throw_statement(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_throw, "`throw`");
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV thrown_code, thrown_arg;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_oppar) { // throw (code) or throw (code, arg)
lex.next();
thrown_code = parse_expr(lex);
if (lex.tok() == tok_comma) {
lex.next();
thrown_arg = parse_expr(lex);
} else {
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
thrown_arg = createV<ast_empty_expression>(loc);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
lex.expect(tok_clpar, "`)`");
} else { // throw code
thrown_code = parse_expr(lex);
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
thrown_arg = createV<ast_empty_expression>(loc);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
lex.expect(tok_semicolon, "`;`");
return createV<ast_throw_statement>(loc, thrown_code, thrown_arg);
}
static AnyV parse_assert_statement(Lexer& lex) {
SrcLocation loc = lex.cur_location();
lex.expect(tok_assert, "`assert`");
lex.expect(tok_oppar, "`(`");
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV cond = parse_expr(lex);
AnyExprV thrown_code;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_comma) { // assert(cond, code)
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
thrown_code = parse_expr(lex);
lex.expect(tok_clpar, "`)`");
} else { // assert(cond) throw code
lex.expect(tok_clpar, "`)`");
lex.expect(tok_throw, "`throw excNo` after assert");
thrown_code = parse_expr(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_semicolon, "`;`");
return createV<ast_assert_statement>(loc, cond, thrown_code);
}
static AnyV parse_try_catch_statement(Lexer& lex) {
SrcLocation loc = lex.cur_location();
lex.expect(tok_try, "`try`");
V<ast_sequence> try_body = parse_sequence(lex);
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
std::vector<AnyExprV> catch_args;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_catch, "`catch`");
SrcLocation catch_loc = lex.cur_location();
if (lex.tok() == tok_oppar) {
lex.next();
catch_args.push_back(parse_catch_variable(lex));
if (lex.tok() == tok_comma) { // catch (excNo, arg)
lex.next();
catch_args.push_back(parse_catch_variable(lex));
} else { // catch (excNo) -> catch (excNo, _)
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
catch_args.push_back(create_catch_underscore_variable(lex));
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
lex.expect(tok_clpar, "`)`");
} else { // catch -> catch (_, _)
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
catch_args.push_back(create_catch_underscore_variable(lex));
catch_args.push_back(create_catch_underscore_variable(lex));
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
}
V<ast_tensor> catch_expr = createV<ast_tensor>(catch_loc, std::move(catch_args));
V<ast_sequence> catch_body = parse_sequence(lex);
return createV<ast_try_catch_statement>(loc, try_body, catch_expr, catch_body);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
AnyV parse_statement(Lexer& lex) {
switch (lex.tok()) {
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
case tok_var: // `var x = 0` is technically an expression, but can not appear in "any place",
case tok_val: // only as a separate declaration
return parse_local_vars_declaration_assignment(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
case tok_opbrace:
return parse_sequence(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
case tok_return:
return parse_return_statement(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
case tok_if:
[Tolk] `bool` type (-1/0 int under the hood) Comparison operators `== / >= /...` return `bool`. Logical operators `&& ||` return bool. Constants `true` and `false` have the `bool` type. Lots of stdlib functions return `bool`, not `int`. Operator `!x` supports both `int` and `bool`. Condition of `if` accepts both `int` and `bool`. Arithmetic operators are restricted to integers. Logical `&&` and `||` accept both `bool` and `int`. No arithmetic operations with bools allowed (only bitwise and logical).
2025-01-13 08:21:24 +00:00
return parse_if_statement(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
case tok_repeat:
return parse_repeat_statement(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
case tok_do:
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
return parse_do_while_statement(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
case tok_while:
return parse_while_statement(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
case tok_throw:
return parse_throw_statement(lex);
case tok_assert:
return parse_assert_statement(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
case tok_try:
return parse_try_catch_statement(lex);
case tok_semicolon: {
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
return createV<ast_empty_statement>(loc);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
case tok_break:
case tok_continue:
lex.error("break/continue from loops are not supported yet");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
default: {
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
AnyExprV expr = parse_expr(lex);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_semicolon, "`;`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
return expr;
}
}
}
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
static AnyV parse_func_body(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
return parse_sequence(lex);
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
static AnyV parse_asm_func_body(Lexer& lex, V<ast_parameter_list> param_list) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_asm, "`asm`");
size_t n_params = param_list->size();
if (n_params > 16) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
throw ParseError{loc, "assembler built-in function can have at most 16 arguments"};
}
std::vector<int> arg_order, ret_order;
if (lex.tok() == tok_oppar) {
lex.next();
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
while (lex.tok() == tok_identifier || lex.tok() == tok_self) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
int arg_idx = param_list->lookup_idx(lex.cur_str());
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
if (arg_idx == -1) {
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
lex.unexpected("parameter name");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
arg_order.push_back(arg_idx);
lex.next();
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_arrow) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
while (lex.tok() == tok_int_const) {
int ret_idx = std::atoi(static_cast<std::string>(lex.cur_str()).c_str());
ret_order.push_back(ret_idx);
lex.next();
}
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_clpar, "`)`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
std::vector<AnyV> asm_commands;
lex.check(tok_string_const, "\"ASM COMMAND\"");
while (lex.tok() == tok_string_const) {
std::string_view asm_command = lex.cur_str();
asm_commands.push_back(createV<ast_string_const>(lex.cur_location(), asm_command, 0));
lex.next();
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_semicolon, "`;`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
return createV<ast_asm_body>(loc, std::move(arg_order), std::move(ret_order), std::move(asm_commands));
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
static AnyV parse_genericsT_list(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
std::vector<AnyV> genericsT_items;
lex.expect(tok_lt, "`<`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
while (true) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.check(tok_identifier, "T");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
std::string_view nameT = lex.cur_str();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
genericsT_items.emplace_back(createV<ast_genericsT_item>(lex.cur_location(), nameT));
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
if (lex.tok() != tok_comma) {
break;
}
lex.next();
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_gt, "`>`");
return createV<ast_genericsT_list>{loc, std::move(genericsT_items)};
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
static V<ast_annotation> parse_annotation(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.check(tok_annotation_at, "`@`");
std::string_view name = lex.cur_str();
AnnotationKind kind = Vertex<ast_annotation>::parse_kind(name);
lex.next();
V<ast_tensor> v_arg = nullptr;
if (lex.tok() == tok_oppar) {
SrcLocation loc_args = lex.cur_location();
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
std::vector<AnyExprV> args;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
args.push_back(parse_expr(lex));
while (lex.tok() == tok_comma) {
lex.next();
args.push_back(parse_expr(lex));
}
lex.expect(tok_clpar, "`)`");
v_arg = createV<ast_tensor>(loc_args, std::move(args));
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
switch (kind) {
case AnnotationKind::unknown:
throw ParseError(loc, "unknown annotation " + static_cast<std::string>(name));
case AnnotationKind::inline_simple:
case AnnotationKind::inline_ref:
case AnnotationKind::pure:
case AnnotationKind::deprecated:
if (v_arg) {
throw ParseError(v_arg->loc, "arguments aren't allowed for " + static_cast<std::string>(name));
}
v_arg = createV<ast_tensor>(loc, {});
break;
case AnnotationKind::method_id:
if (!v_arg || v_arg->size() != 1 || v_arg->get_item(0)->type != ast_int_const) {
throw ParseError(loc, "expecting `(number)` after " + static_cast<std::string>(name));
}
break;
}
return createV<ast_annotation>(loc, kind, v_arg);
}
static AnyV parse_function_declaration(Lexer& lex, const std::vector<V<ast_annotation>>& annotations) {
SrcLocation loc = lex.cur_location();
bool is_get_method = lex.tok() == tok_get;
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (is_get_method && lex.tok() == tok_fun) {
lex.next(); // 'get f()' and 'get fun f()' both correct
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.check(tok_identifier, "function name identifier");
std::string_view f_name = lex.cur_str();
bool is_entrypoint =
f_name == "main" || f_name == "onInternalMessage" || f_name == "onExternalMessage" ||
f_name == "onRunTickTock" || f_name == "onSplitPrepare" || f_name == "onSplitInstall";
bool is_FunC_entrypoint =
f_name == "recv_internal" || f_name == "recv_external" ||
f_name == "run_ticktock" || f_name == "split_prepare" || f_name == "split_install";
if (is_FunC_entrypoint) {
lex.error("this is a reserved FunC/Fift identifier; you need `onInternalMessage`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
auto v_ident = createV<ast_identifier>(lex.cur_location(), f_name);
lex.next();
V<ast_genericsT_list> genericsT_list = nullptr;
if (lex.tok() == tok_lt) { // 'fun f<T1,T2>'
genericsT_list = parse_genericsT_list(lex)->as<ast_genericsT_list>();
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
V<ast_parameter_list> v_param_list = parse_parameter_list(lex)->as<ast_parameter_list>();
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
bool accepts_self = !v_param_list->empty() && v_param_list->get_param(0)->param_name == "self";
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
int n_mutate_params = v_param_list->get_mutate_params_count();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
TypePtr ret_type = nullptr;
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
bool returns_self = false;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (lex.tok() == tok_colon) { // : <ret_type> (if absent, it means "auto infer", not void)
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
if (lex.tok() == tok_self) {
if (!accepts_self) {
lex.error("only a member function can return `self` (which accepts `self` first parameter)");
}
lex.next();
returns_self = true;
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
ret_type = TypeDataVoid::create();
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
} else {
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
ret_type = parse_type_from_tokens(lex);
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
}
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
if (is_entrypoint && (is_get_method || genericsT_list || n_mutate_params || accepts_self)) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
throw ParseError(loc, "invalid declaration of a reserved function");
}
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
if (is_get_method && (genericsT_list || n_mutate_params || accepts_self)) {
throw ParseError(loc, "get methods can't have `mutate` and `self` params");
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
AnyV v_body = nullptr;
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
if (lex.tok() == tok_builtin) {
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
v_body = createV<ast_empty_statement>(lex.cur_location());
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_semicolon, "`;`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
} else if (lex.tok() == tok_opbrace) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
v_body = parse_func_body(lex);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
} else if (lex.tok() == tok_asm) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
if (!ret_type) {
lex.error("asm function must specify return type");
}
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
v_body = parse_asm_func_body(lex, v_param_list);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
} else {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.unexpected("{ function body }");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
int flags = 0;
if (is_entrypoint) {
flags |= FunctionData::flagIsEntrypoint;
}
if (is_get_method) {
flags |= FunctionData::flagGetMethod;
}
if (accepts_self) {
flags |= FunctionData::flagAcceptsSelf;
}
if (returns_self) {
flags |= FunctionData::flagReturnsSelf;
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
td::RefInt256 method_id;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
for (auto v_annotation : annotations) {
switch (v_annotation->kind) {
case AnnotationKind::inline_simple:
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
flags |= FunctionData::flagInline;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
break;
case AnnotationKind::inline_ref:
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
flags |= FunctionData::flagInlineRef;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
break;
case AnnotationKind::pure:
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
flags |= FunctionData::flagMarkedAsPure;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
break;
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
case AnnotationKind::method_id: {
[Tolk] Get rid of ~tilda with `mutate` and `self` methods This is a very big change. If FunC has `.methods()` and `~methods()`, Tolk has only dot, one and only way to call a `.method()`. A method may mutate an object, or may not. It's a behavioral and semantic difference from FunC. - `cs.loadInt(32)` modifies a slice and returns an integer - `b.storeInt(x, 32)` modifies a builder - `b = b.storeInt()` also works, since it not only modifies, but returns - chained methods also work, they return `self` - everything works exactly as expected, similar to JS - no runtime overhead, exactly same Fift instructions - custom methods are created with ease - tilda `~` does not exist in Tolk at all
2024-10-31 07:18:54 +00:00
if (is_get_method || genericsT_list || is_entrypoint || n_mutate_params || accepts_self) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
v_annotation->error("@method_id can be specified only for regular functions");
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
auto v_int = v_annotation->get_arg()->get_item(0)->as<ast_int_const>();
if (v_int->intval.is_null() || !v_int->intval->signed_fits_bits(32)) {
v_int->error("invalid integer constant");
}
method_id = v_int->intval;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
break;
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
case AnnotationKind::deprecated:
// no special handling
break;
default:
v_annotation->error("this annotation is not applicable to functions");
}
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
return createV<ast_function_declaration>(loc, v_ident, v_param_list, v_body, ret_type, genericsT_list, std::move(method_id), flags);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
static AnyV parse_tolk_required_version(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.next_special(tok_semver, "semver"); // syntax: "tolk 0.6"
std::string semver = static_cast<std::string>(lex.cur_str());
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
// for simplicity, there is no syntax ">= version" and so on, just strict compare
if (TOLK_VERSION != semver && TOLK_VERSION != semver + ".0") { // 0.6 = 0.6.0
loc.show_warning("the contract is written in Tolk v" + semver + ", but you use Tolk compiler v" + TOLK_VERSION + "; probably, it will lead to compilation errors or hash changes");
}
[Tolk] AST-based semantic analysis, get rid of Expr This is a huge refactoring focusing on untangling compiler internals (previously forked from FunC). The goal is to convert AST directly to Op (a kind of IR representation), doing all code analysis at AST level. Noteable changes: - AST-based semantic kernel includes: registering global symbols, scope handling and resolving local/global identifiers, lvalue/rvalue calc and check, implicit return detection, mutability analysis, pure/impure validity checks, simple constant folding - values of `const` variables are calculated NOT based on CodeBlob, but via a newly-introduced AST-based constant evaluator - AST vertices are now inherited from expression/statement/other; expression vertices have common properties (TypeExpr, lvalue/rvalue) - symbol table is rewritten completely, SymDef/SymVal no longer exist, lexer now doesn't need to register identifiers - AST vertices have references to symbols, filled at different stages of pipeline - the remaining "FunC legacy part" is almost unchanged besides Expr which was fully dropped; AST is converted to Ops (IR) directly
2024-12-16 18:19:45 +00:00
return createV<ast_tolk_required_version>(loc, semver); // semicolon is not necessary
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
static AnyV parse_import_directive(Lexer& lex) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
SrcLocation loc = lex.cur_location();
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
lex.expect(tok_import, "`import`");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.check(tok_string_const, "source file name");
std::string_view rel_filename = lex.cur_str();
if (rel_filename.empty()) {
lex.error("imported file name is an empty string");
}
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
auto v_str = createV<ast_string_const>(lex.cur_location(), rel_filename, 0);
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
lex.next();
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
return createV<ast_import_directive>(loc, v_str); // semicolon is not necessary
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
// the main (exported) function
[Tolk] Compilation pipeline, register global symbols in advance Since I've implemented AST, now I can drop forward declarations. Instead, I traverse AST of all files and register global symbols (functions, constants, global vars) as a separate step, in advance. That's why, while converting AST to Expr/Op, all available symbols are already registered. This greatly simplifies "intermediate state" of yet unknown functions and checking them afterward. Redeclaration of local variables (inside the same scope) is now also prohibited.
2024-10-31 07:04:58 +00:00
AnyV parse_src_file_to_ast(const SrcFile* file) {
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
std::vector<AnyV> toplevel_declarations;
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
std::vector<V<ast_annotation>> annotations;
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
Lexer lex(file);
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
while (!lex.is_eof()) {
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
switch (lex.tok()) {
case tok_tolk:
if (!annotations.empty()) {
lex.unexpected("declaration after @annotations");
}
toplevel_declarations.push_back(parse_tolk_required_version(lex));
break;
case tok_import:
if (!annotations.empty()) {
lex.unexpected("declaration after @annotations");
}
[Tolk] Rewrite the type system from Hindley-Milner to static typing FunC's (and Tolk's before this PR) type system is based on Hindley-Milner. This is a common approach for functional languages, where types are inferred from usage through unification. As a result, type declarations are not necessary: () f(a,b) { return a+b; } // a and b now int, since `+` (int, int) While this approach works for now, problems arise with the introduction of new types like bool, where `!x` must handle both int and bool. It will also become incompatible with int32 and other strict integers. This will clash with structure methods, struggle with proper generics, and become entirely impractical for union types. This PR completely rewrites the type system targeting the future. 1) type of any expression is inferred and never changed 2) this is available because dependent expressions already inferred 3) forall completely removed, generic functions introduced (they work like template functions actually, instantiated while inferring) 4) instantiation `<...>` syntax, example: `t.tupleAt<int>(0)` 5) `as` keyword, for example `t.tupleAt(0) as int` 6) methods binding is done along with type inferring, not before ("before", as worked previously, was always a wrong approach)
2024-12-30 15:31:27 +00:00
toplevel_declarations.push_back(parse_import_directive(lex));
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
break;
case tok_semicolon:
if (!annotations.empty()) {
lex.unexpected("declaration after @annotations");
}
lex.next(); // don't add ast_empty, no need
break;
case tok_annotation_at:
annotations.push_back(parse_annotation(lex));
break;
case tok_global:
toplevel_declarations.push_back(parse_global_var_declaration(lex, annotations));
annotations.clear();
break;
case tok_const:
toplevel_declarations.push_back(parse_constant_declaration(lex, annotations));
annotations.clear();
break;
case tok_fun:
case tok_get:
toplevel_declarations.push_back(parse_function_declaration(lex, annotations));
annotations.clear();
break;
case tok_export:
case tok_struct:
case tok_enum:
case tok_operator:
case tok_infix:
lex.error("`" + static_cast<std::string>(lex.cur_str()) +"` is not supported yet");
default:
lex.unexpected("fun or get");
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
}
}
[Tolk] v0.6 syntax: `fun`, `import`, `var`, types on the right, etc. Lots of changes, actually. Most noticeable are: - traditional //comments - #include -> import - a rule "import what you use" - ~ found -> !found (for -1/0) - null() -> null - is_null?(v) -> v == null - throw is a keyword - catch with swapped arguments - throw_if, throw_unless -> assert - do until -> do while - elseif -> else if - drop ifnot, elseifnot - drop rarely used operators A testing framework also appears here. All tests existed earlier, but due to significant syntax changes, their history is useless.
2024-10-31 07:11:41 +00:00
[Tolk] Implement AST: intermediate representation of tolk files Now, the whole .tolk file can be loaded as AST tree and then converted to Expr/Op. This gives a great ability to implement AST transformations. In the future, more and more code analysis will be moved out of legacy to AST-level.
2024-10-31 07:03:33 +00:00
return createV<ast_tolk_file>(file, std::move(toplevel_declarations));
}
} // namespace tolk
Reference in a new issue Copy permalink