[Tolk] Tolk v0.5.0 as FunC v0.5.0 could have been like

All changes from PR "FunC v0.5.0": #1026 Instead of developing FunC, we decided to fork it. BTW, the first Tolk release will be v0.6, a metaphor of FunC v0.5 that missed a chance to occur.
ton-blockchain · Nov 1, 2024 · ebbab54 · ebbab54
1 parent 82648eb
commit ebbab54
Show file tree

Hide file tree

Showing 21 changed files with 1,362 additions and 806 deletions.
diff --git a/crypto/smartcont/mathlib.tolk b/crypto/smartcont/mathlib.tolk
@@ -81,14 +81,14 @@ int fixed248::acot(int x) inline_ref;
 
 ;; random number uniformly distributed in [0..1)
 ;; fixed248 random();
-int fixed248::random() impure inline;
+int fixed248::random() inline;
 ;; random number with standard normal distribution (2100 gas on average)
 ;; fixed248 nrand();
-int fixed248::nrand() impure inline;
+int fixed248::nrand() inline;
 ;; generates a random number approximately distributed according to the standard normal distribution (1200 gas)
 ;; (fails chi-squared test, but it is shorter and faster than fixed248::nrand())
 ;; fixed248 nrand_fast();
-int fixed248::nrand_fast() impure inline;
+int fixed248::nrand_fast() inline;
 
 -}  ;; end (declarations)
 
@@ -880,7 +880,7 @@ int fixed248::acot(int x) inline_ref {
 ;; generated by Kinderman--Monahan ratio method modified by J.Leva
 ;; spends ~ 2k..3k gas on average
 ;; fixed252 nrand();
-int nrand_f252() impure inline_ref {
+int nrand_f252() inline_ref {
   var (x, s, t, A, B, r0) = (nan(), touch(29483) << 236, touch(-3167) << 239, 12845, 16693, 9043);
   ;; 4/sqrt(e*Pi) = 1.369 loop iterations on average
   do {
@@ -910,7 +910,7 @@ int nrand_f252() impure inline_ref {
 ;; generates a random number approximately distributed according to the standard normal distribution
 ;; much faster than nrand_f252(), should be suitable for most purposes when only several random numbers are needed
 ;; fixed252 nrand_fast();
-int nrand_fast_f252() impure inline_ref {
+int nrand_fast_f252() inline_ref {
   int t = touch(-3) << 253;   ;; -6. as fixed252
   repeat (12) {
     t += random() / 16;         ;; add together 12 uniformly random numbers
@@ -920,18 +920,18 @@ int nrand_fast_f252() impure inline_ref {
 
 ;; random number uniformly distributed in [0..1)
 ;; fixed248 random();
-int fixed248::random() impure inline {
+int fixed248::random() inline {
   return random() >> 8;
 }
 
 ;; random number with standard normal distribution
 ;; fixed248 nrand();
-int fixed248::nrand() impure inline {
+int fixed248::nrand() inline {
   return nrand_f252() ~>> 4;
 }
 
 ;; generates a random number approximately distributed according to the standard normal distribution
 ;; fixed248 nrand_fast();
-int fixed248::nrand_fast() impure inline {
+int fixed248::nrand_fast() inline {
   return nrand_fast_f252() ~>> 4;
 }
diff --git a/crypto/smartcont/stdlib.tolk b/crypto/smartcont/stdlib.tolk
diff --git a/lite-client/lite-client.cpp b/lite-client/lite-client.cpp
@@ -926,7 +926,7 @@ bool TestNode::show_help(std::string command) {
          "saveaccount[code|data] <filename> <addr> [<block-id-ext>]\tSaves into specified file the most recent state "
          "(StateInit) or just the code or data of specified account; <addr> is in "
          "[<workchain>:]<hex-or-base64-addr> format\n"
-         "runmethod[full] <addr> [<block-id-ext>] <method-id> <params>...\tRuns GET method <method-id> of account "
+         "runmethod[full] <addr> [<block-id-ext>] <name> <params>...\tRuns GET method <name> of account "
          "<addr> "
          "with specified parameters\n"
          "dnsresolve [<block-id-ext>] <domain> [<category>]\tResolves a domain starting from root dns smart contract\n"

diff --git a/tolk/CMakeLists.txt b/tolk/CMakeLists.txt
@@ -24,6 +24,10 @@ target_link_libraries(tolk PUBLIC git ton_crypto)  # todo replace with ton_crypt
 if (WINGETOPT_FOUND)
   target_link_libraries_system(tolk wingetopt)
 endif ()
+if (${TOLK_DEBUG}) # -DTOLK_DEBUG=1 in CMake options => #define TOLK_DEBUG (for development purposes)
+  message(STATUS "TOLK_DEBUG is ON")
+  target_compile_definitions(tolk PRIVATE TOLK_DEBUG=1)
+endif()
 
 if (USE_EMSCRIPTEN)
   add_executable(tolkfiftlib tolk-wasm.cpp ${TOLK_SOURCE})

diff --git a/tolk/abscode.cpp b/tolk/abscode.cpp
@@ -221,15 +221,6 @@ void VarDescrList::show(std::ostream& os) const {
   os << " ]\n";
 }
 
-void Op::flags_set_clear(int set, int clear) {
-  flags = (flags | set) & ~clear;
-  for (auto& op : block0) {
-    op.flags_set_clear(set, clear);
-  }
-  for (auto& op : block1) {
-    op.flags_set_clear(set, clear);
-  }
-}
 void Op::split_vars(const std::vector<TmpVar>& vars) {
   split_var_list(left, vars);
   split_var_list(right, vars);
@@ -294,7 +285,7 @@ void Op::show(std::ostream& os, const std::vector<TmpVar>& vars, std::string pfx
   if (noreturn()) {
     dis += "<noret> ";
   }
-  if (!is_pure()) {
+  if (impure()) {
     dis += "<impure> ";
   }
   switch (cl) {
@@ -467,12 +458,6 @@ void Op::show_block(std::ostream& os, const Op* block, const std::vector<TmpVar>
   os << pfx << "}";
 }
 
-void CodeBlob::flags_set_clear(int set, int clear) {
-  for (auto& op : ops) {
-    op.flags_set_clear(set, clear);
-  }
-}
-
 std::ostream& operator<<(std::ostream& os, const CodeBlob& code) {
   code.print(os);
   return os;

diff --git a/tolk/analyzer.cpp b/tolk/analyzer.cpp
@@ -360,10 +360,10 @@ bool Op::compute_used_vars(const CodeBlob& code, bool edit) {
     case _Tuple:
     case _UnTuple: {
       // left = EXEC right;
-      if (!next_var_info.count_used(left) && is_pure()) {
+      if (!next_var_info.count_used(left) && !impure()) {
         // all variables in `left` are not needed
         if (edit) {
-          disable();
+          set_disabled();
         }
         return std_compute_used_vars(true);
       }
@@ -372,7 +372,7 @@ bool Op::compute_used_vars(const CodeBlob& code, bool edit) {
     case _SetGlob: {
       // GLOB = right
       if (right.empty() && edit) {
-        disable();
+        set_disabled();
       }
       return std_compute_used_vars(right.empty());
     }
@@ -399,7 +399,7 @@ bool Op::compute_used_vars(const CodeBlob& code, bool edit) {
       }
       if (!cnt && edit) {
         // all variables in `left` are not needed
-        disable();
+        set_disabled();
       }
       return set_var_info(std::move(new_var_info));
     }
@@ -860,15 +860,45 @@ VarDescrList Op::fwd_analyze(VarDescrList values) {
   }
 }
 
-bool Op::set_noreturn(bool nr) {
-  if (nr) {
+void Op::set_disabled(bool flag) {
+  if (flag) {
+    flags |= _Disabled;
+  } else {
+    flags &= ~_Disabled;
+  }
+}
+
+
+bool Op::set_noreturn(bool flag) {
+  if (flag) {
     flags |= _NoReturn;
   } else {
     flags &= ~_NoReturn;
   }
-  return nr;
+  return flag;
+}
+
+void Op::set_impure(const CodeBlob &code) {
+  // todo calling this function with `code` is a bad design (flags are assigned after Op is constructed)
+  // later it's better to check this somewhere in code.emplace_back()
+  if (code.flags & CodeBlob::_ForbidImpure) {
+    throw ParseError(where, "An impure operation in a pure function");
+  }
+  flags |= _Impure;
 }
 
+void Op::set_impure(const CodeBlob &code, bool flag) {
+  if (flag) {
+    if (code.flags & CodeBlob::_ForbidImpure) {
+      throw ParseError(where, "An impure operation in a pure function");
+    }
+    flags |= _Impure;
+  } else {
+    flags &= ~_Impure;
+  }
+}
+
+
 bool Op::mark_noreturn() {
   switch (cl) {
     case _Nop:
@@ -888,13 +918,14 @@ bool Op::mark_noreturn() {
     case _Call:
       return set_noreturn(next->mark_noreturn());
     case _Return:
-      return set_noreturn(true);
+      return set_noreturn();
     case _If:
     case _TryCatch:
+      // note, that & | (not && ||) here and below is mandatory to invoke both left and right calls
       return set_noreturn((static_cast<int>(block0->mark_noreturn()) & static_cast<int>(block1 && block1->mark_noreturn())) | static_cast<int>(next->mark_noreturn()));
     case _Again:
       block0->mark_noreturn();
-      return set_noreturn(true);
+      return set_noreturn();
     case _Until:
       return set_noreturn(static_cast<int>(block0->mark_noreturn()) | static_cast<int>(next->mark_noreturn()));
     case _While:

diff --git a/tolk/asmops.cpp b/tolk/asmops.cpp
@@ -317,6 +317,9 @@ void AsmOpList::show_var_ext(std::ostream& os, std::pair<var_idx_t, const_idx_t>
     os << '_' << i;
   } else {
     var_names_->at(i).show(os, 2);
+    // if (!var_names_->at(i).v_type->is_int()) {
+    //   os << '<'; var_names_->at(i).v_type->print(os); os << '>';
+    // }
   }
   if ((unsigned)j < constants_.size() && constants_[j].not_null()) {
     os << '=' << constants_[j];

diff --git a/tolk/builtins.cpp b/tolk/builtins.cpp
@@ -26,10 +26,10 @@ using namespace std::literals::string_literals;
  */
 
 int glob_func_cnt, undef_func_cnt, glob_var_cnt, const_cnt;
-std::vector<SymDef*> glob_func, glob_vars;
+std::vector<SymDef*> glob_func, glob_vars, glob_get_methods;
 std::set<std::string> prohibited_var_names;
 
-SymDef* predefine_builtin_func(std::string name, TypeExpr* func_type) {
+SymDef* define_builtin_func_impl(const std::string& name, SymValAsmFunc* func_val) {
   if (name.back() == '_') {
     prohibited_var_names.insert(name);
   }
@@ -42,30 +42,40 @@ SymDef* predefine_builtin_func(std::string name, TypeExpr* func_type) {
     std::cerr << "fatal: global function `" << name << "` already defined" << std::endl;
     std::exit(1);
   }
+  func_val->flags |= SymValFunc::flagBuiltinFunction;
+  def->value = func_val;
+#ifdef TOLK_DEBUG
+  dynamic_cast<SymValAsmFunc*>(def->value)->name = name;
+#endif
   return def;
 }
 
-template <typename T>
-SymDef* define_builtin_func(std::string name, TypeExpr* func_type, const T& func, bool impure = false) {
-  SymDef* def = predefine_builtin_func(name, func_type);
-  def->value = new SymValAsmFunc{func_type, func, impure};
-  return def;
+SymDef* define_builtin_func(const std::string& name, TypeExpr* func_type, const simple_compile_func_t& func, bool impure = false) {
+  return define_builtin_func_impl(name, new SymValAsmFunc{func_type, func, !impure});
+}
+
+SymDef* define_builtin_func(const std::string& name, TypeExpr* func_type, const compile_func_t& func, bool impure = false) {
+  return define_builtin_func_impl(name, new SymValAsmFunc{func_type, func, !impure});
+}
+
+SymDef* define_builtin_func(const std::string& name, TypeExpr* func_type, const AsmOp& macro, bool impure = false) {
+  return define_builtin_func_impl(name, new SymValAsmFunc{func_type, make_simple_compile(macro), !impure});
 }
 
-template <typename T>
-SymDef* define_builtin_func(std::string name, TypeExpr* func_type, const T& func, std::initializer_list<int> arg_order,
+SymDef* define_builtin_func(const std::string& name, TypeExpr* func_type, const simple_compile_func_t& func, std::initializer_list<int> arg_order,
                             std::initializer_list<int> ret_order = {}, bool impure = false) {
-  SymDef* def = predefine_builtin_func(name, func_type);
-  def->value = new SymValAsmFunc{func_type, func, arg_order, ret_order, impure};
-  return def;
+  return define_builtin_func_impl(name, new SymValAsmFunc{func_type, func, arg_order, ret_order, !impure});
 }
 
-SymDef* define_builtin_func(std::string name, TypeExpr* func_type, const AsmOp& macro,
+SymDef* define_builtin_func(const std::string& name, TypeExpr* func_type, const compile_func_t& func, std::initializer_list<int> arg_order,
+                            std::initializer_list<int> ret_order = {}, bool impure = false) {
+  return define_builtin_func_impl(name, new SymValAsmFunc{func_type, func, arg_order, ret_order, !impure});
+}
+
+SymDef* define_builtin_func(const std::string& name, TypeExpr* func_type, const AsmOp& macro,
                             std::initializer_list<int> arg_order, std::initializer_list<int> ret_order = {},
                             bool impure = false) {
-  SymDef* def = predefine_builtin_func(name, func_type);
-  def->value = new SymValAsmFunc{func_type, make_simple_compile(macro), arg_order, ret_order, impure};
-  return def;
+  return define_builtin_func_impl(name, new SymValAsmFunc{func_type, make_simple_compile(macro), arg_order, ret_order, !impure});
 }
 
 SymDef* force_autoapply(SymDef* def) {
@@ -262,7 +272,7 @@ int emulate_lshift(int a, int b) {
   }
   int t = ((b & VarDescr::_NonZero) ? VarDescr::_Even : 0);
   t |= b & VarDescr::_Finite;
-  return emulate_mul(a, VarDescr::_Int | VarDescr::_Pos | VarDescr::_NonZero | VarDescr::_Even | t);
+  return emulate_mul(a, VarDescr::_Int | VarDescr::_Pos | VarDescr::_NonZero | t);
 }
 
 int emulate_div(int a, int b) {
@@ -308,7 +318,7 @@ int emulate_rshift(int a, int b) {
   }
   int t = ((b & VarDescr::_NonZero) ? VarDescr::_Even : 0);
   t |= b & VarDescr::_Finite;
-  return emulate_div(a, VarDescr::_Int | VarDescr::_Pos | VarDescr::_NonZero | VarDescr::_Even | t);
+  return emulate_div(a, VarDescr::_Int | VarDescr::_Pos | VarDescr::_NonZero | t);
 }
 
 int emulate_mod(int a, int b, int round_mode = -1) {
@@ -1128,9 +1138,9 @@ void define_builtins() {
   auto Int3 = TypeExpr::new_tensor({Int, Int, Int});
   auto TupleInt = TypeExpr::new_tensor({Tuple, Int});
   auto SliceInt = TypeExpr::new_tensor({Slice, Int});
-  auto X = TypeExpr::new_var();
-  auto Y = TypeExpr::new_var();
-  auto Z = TypeExpr::new_var();
+  auto X = TypeExpr::new_var(0);
+  auto Y = TypeExpr::new_var(1);
+  auto Z = TypeExpr::new_var(2);
   auto XY = TypeExpr::new_tensor({X, Y});
   auto arith_bin_op = TypeExpr::new_map(Int2, Int);
   auto arith_un_op = TypeExpr::new_map(Int, Int);

diff --git a/tolk/codegen.cpp b/tolk/codegen.cpp
@@ -437,6 +437,7 @@ bool Op::generate_code_step(Stack& stack) {
       if (disabled()) {
         return true;
       }
+      // fun_ref can be nullptr for Op::_CallInd (invoke a variable, not a function)
       SymValFunc* func = (fun_ref ? dynamic_cast<SymValFunc*>(fun_ref->value) : nullptr);
       auto arg_order = (func ? func->get_arg_order() : nullptr);
       auto ret_order = (func ? func->get_ret_order() : nullptr);
@@ -486,27 +487,24 @@ bool Op::generate_code_step(Stack& stack) {
       };
       if (cl == _CallInd) {
         exec_callxargs((int)right.size() - 1, (int)left.size());
+      } else if (auto asm_fv = dynamic_cast<const SymValAsmFunc*>(fun_ref->value)) {
+        std::vector<VarDescr> res;
+        res.reserve(left.size());
+        for (var_idx_t i : left) {
+          res.emplace_back(i);
+        }
+        asm_fv->compile(stack.o, res, args, where);  // compile res := f (args)
       } else {
-        auto func = dynamic_cast<const SymValAsmFunc*>(fun_ref->value);
-        if (func) {
-          std::vector<VarDescr> res;
-          res.reserve(left.size());
-          for (var_idx_t i : left) {
-            res.emplace_back(i);
-          }
-          func->compile(stack.o, res, args, where);  // compile res := f (args)
+        auto fv = dynamic_cast<const SymValCodeFunc*>(fun_ref->value);
+        // todo can be fv == nullptr?
+        std::string name = symbols.get_name(fun_ref->sym_idx);
+        if (fv && (fv->is_inline() || fv->is_inline_ref())) {
+          stack.o << AsmOp::Custom(name + " INLINECALLDICT", (int)right.size(), (int)left.size());
+        } else if (fv && fv->code && fv->code->require_callxargs) {
+          stack.o << AsmOp::Custom(name + (" PREPAREDICT"), 0, 2);
+          exec_callxargs((int)right.size() + 1, (int)left.size());
         } else {
-          auto fv = dynamic_cast<const SymValCodeFunc*>(fun_ref->value);
-          std::string name = symbols.get_name(fun_ref->sym_idx);
-          bool is_inline = (fv && (fv->flags & 3));
-          if (is_inline) {
-            stack.o << AsmOp::Custom(name + " INLINECALLDICT", (int)right.size(), (int)left.size());
-          } else if (fv && fv->code && fv->code->require_callxargs) {
-            stack.o << AsmOp::Custom(name + (" PREPAREDICT"), 0, 2);
-            exec_callxargs((int)right.size() + 1, (int)left.size());
-          } else {
-            stack.o << AsmOp::Custom(name + " CALLDICT", (int)right.size(), (int)left.size());
-          }
+          stack.o << AsmOp::Custom(name + " CALLDICT", (int)right.size(), (int)left.size());
         }
       }
       stack.s.resize(k);