Should this example compile at this stage of Blaise Development ? If it should, is there a deeper issue ?

[Schmitty2005]

I understand that blaise is not ready, but I wanted to point this out in case there was a deeper issue going on. Looking at the code I would think this should work, as there are functions with results in the compiler code itself.

I have the following code :

program addInt;

function myAdd ( x : integer; y : integer ) : integer;
begin
  result := x + y;
end;

var
  z : integer;
begin
  z:= myAdd (2, 3);
end.

I get this error addint.ssa:33: invalid type for second operand %_var_result in storew

Here is the addint.ssa :

# Generated by Blaise Compiler
# Source: addInt


export data $z = { w 0 }

data $__cn_TObject = { w -1, w 7, w 7, b "TObject", b 0 }
data $typeinfo_TObject = { l 0, l 0, l $__cn_TObject + 12, l 0, l 8, l $_FieldCleanup_TObject, l $vtable_TObject }

data $vtable_TObject = { l $typeinfo_TObject, l $TObject_Destroy, l $TObject_ToString }

function $_FieldCleanup_TObject(l %self) {
@start
  ret
}

export function w $myAdd(w %_par_x, w %_par_y) {
@start# Generated by Blaise Compiler
# Source: addInt


export data $z = { w 0 }

data $__cn_TObject = { w -1, w 7, w 7, b "TObject", b 0 }
data $typeinfo_TObject = { l 0, l 0, l $__cn_TObject + 12, l 0, l 8, l $_FieldCleanup_TObject, l $vtable_TObject }

data $vtable_TObject = { l $typeinfo_TObject, l $TObject_Destroy, l $TObject_ToString }

function $_FieldCleanup_TObject(l %self) {
@start
  ret
}

export function w $myAdd(w %_par_x, w %_par_y) {
@start
  %_var_x =l alloc4 1
  storew %_par_x, %_var_x
  %_var_y =l alloc4 1
  storew %_par_y, %_var_y
  %_var_Result =l alloc4 1
  storew 0, %_var_Result
  %_t0 =w loadw %_var_x
  %_t1 =w loadw %_var_y
  %_t2 =w add %_t0, %_t1
  storew %_t2, %_var_result
  jmp @func_exit_0
@func_exit_0
  %_t3 =w loadw %_var_Result
  ret %_t3
}

export function w $main(w %argc, l %argv) {
@start
  call $_SetArgs(w %argc, l %argv)
  %_t4 =w copy 2
  %_t5 =w copy 3
  %_t6 =w call $myAdd(w %_t4, w %_t5)
  storew %_t6, $z
  jmp @main_exit
@main_exit
  ret 0
}

  %_var_x =l alloc4 1
  storew %_par_x, %_var_x
  %_var_y =l alloc4 1
  storew %_par_y, %_var_y
  %_var_Result =l alloc4 1
  storew 0, %_var_Result
  %_t0 =w loadw %_var_x
  %_t1 =w loadw %_var_y
  %_t2 =w add %_t0, %_t1
  storew %_t2, %_var_result
  jmp @func_exit_0
@func_exit_0
  %_t3 =w loadw %_var_Result
  ret %_t3
}

export function w $main(w %argc, l %argv) {
@start
  call $_SetArgs(w %argc, l %argv)
  %_t4 =w copy 2
  %_t5 =w copy 3
  %_t6 =w call $myAdd(w %_t4, w %_t5)
  storew %_t6, $z
  jmp @main_exit
@main_exit
  ret 0
}

[andrewd207]

Without having any idea of the cause or fix, storew %_t2, %_var_result is using a lowercase "r" instead of all the other references to it. Seems to be the problem :) Changing your pascal code to use "Result" fixes the problem

[Schmitty2005]

Ahh.... a minor Blaise issue then. Thank you!

[Schmitty2005]

@graemeg here is a small bug 😃

[andrewd207]

fwiw I dug into this a bit. This fixes it. I don't have a good grasp of the big picture so I don't know if this is the correct fix.
(and Graeme is a busy guy who doesn't want a million patches at the moment :)
On further testing referring to 'x' as 'X' has the same problem.

a few lines above it has VarSym := FTable.Lookup(AAssign.Name); which does a case insensitive search for the variable.

diff --git a/compiler/src/main/pascal/uSemantic.pas b/compiler/src/main/pascal/uSemantic.pas
index 3fd0d14..433812b 100644
--- a/compiler/src/main/pascal/uSemantic.pas
+++ b/compiler/src/main/pascal/uSemantic.pas
@@ -3328,6 +3328,7 @@ begin
       Format('''%s'' is not a variable', [AAssign.Name]),
       AAssign.Line, AAssign.Col);
 
+  AAssign.Name            := VarSym.Name;
   AAssign.IsVarParam      := (VarSym.Kind = skVarParameter);
   AAssign.ResolvedLhsType := VarSym.TypeDesc;
   AAssign.IsWeakLhs       := VarSym.IsWeak;

[andrewd207]

I imagine 95% of the time the variable is typed the same way as it's declared so 95% of the time this is redundant.

[graemeg]

Thank you both. The first community bug report. 😄 Fixed in commit f2e78bd fix(semantic): normalise identifier casing to declared symbol name. I also found another location where that had to be fixed.

[Schmitty2005]

Working here ! Thank you!

[andrewd207]

The real question is how did you manage to write so much code without one time using a variable differently (case) 🤣
Glad if it helped :)