When resolving a string literal we examine each character against low
and high bounds of the expected type. We stored each character as an Int
and implicitly converted it to Uint twice: for "<" and ">" operators.
Now we store convert it to Uint explicitly and only once.
Cleanup related to handling of compile-time constraints errors.
Semantics is unaffacted.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* sem_res.adb (Resolve_String_Literal): Avoid unnecessary
conversions inside "<" and ">" bodies.diff --git a/gcc/ada/sem_res.adb b/gcc/ada/sem_res.adb
--- a/gcc/ada/sem_res.adb
+++ b/gcc/ada/sem_res.adb
@@ -11722,14 +11722,14 @@ package body Sem_Res is
Comp_Typ_Hi : constant Node_Id :=
Type_High_Bound (Component_Type (Typ));
- Char_Val : Int;
+ Char_Val : Uint;
begin
if Compile_Time_Known_Value (Comp_Typ_Lo)
and then Compile_Time_Known_Value (Comp_Typ_Hi)
then
for J in 1 .. Strlen loop
- Char_Val := Int (Get_String_Char (Str, J));
+ Char_Val := UI_From_CC (Get_String_Char (Str, J));
if Char_Val < Expr_Value (Comp_Typ_Lo)
or else Char_Val > Expr_Value (Comp_Typ_Hi)
