REPL str support
#2724
REPL str support
#2724
Conversation
| if (y == NULL) { | ||
| *x = NULL; | ||
| return; | ||
| } |
There was a problem hiding this comment.
Why do we need this? Could you share an example where this fails currently as of the main branch?
There was a problem hiding this comment.
Without the above change, the following fails (SEGFAULT):
❯ lp
>>> s: str
>>> s
The stack address was not found in any shared library or the main program, the stack is probably corrupted. Aborting.Tested by Test Case PythonCompiler string 2.
There was a problem hiding this comment.
I think it is fine. I am curious about the following:
- why does
_lfortran_strcpyget called in this case? - are you able to reproduce this same bug in normal/regular
lpythoninvocation?
There was a problem hiding this comment.
why does
_lfortran_strcpyget called in this case?
The global_stmt function that is constructed with the top-level expressions, copies the string into the return variable. You can run the example code above with lp -v flag to look at the ASR and LLVM IR generated.
The ASR and IR for the second line:
(TranslationUnit
(SymbolTable
1
{
__main__:
(Module
(SymbolTable
2
{
__main__global_stmts_2__:
(Function
(SymbolTable
3
{
__main__global_stmts_2__1:
(Variable
3
__main__global_stmts_2__1
[]
ReturnVar
()
()
Default
(Character 1 -2 ())
()
BindC
Public
Required
.false.
)
})
__main__global_stmts_2__
(FunctionType
[]
(Character 1 -2 ())
BindC
Implementation
()
.false.
.false.
.false.
.false.
.false.
[]
.false.
)
[]
[]
[(Assignment
(Var 3 __main__global_stmts_2__1)
(Var 2 s)
()
)]
(Var 3 __main__global_stmts_2__1)
Public
.false.
.false.
()
),
s:
(Variable
2
s
[]
Local
()
()
Default
(Character 1 -2 ())
()
Interactive
Public
Required
.false.
)
})
__main__
[]
.false.
.false.
)
})
[]
); ModuleID = 'LFortran'
source_filename = "LFortran"
@s = external global i8*
define i8* @__main__global_stmts_2__() {
.entry:
%__main__global_stmts_2__1 = alloca i8*, align 8
store i8* null, i8** %__main__global_stmts_2__1, align 8
%0 = load i8*, i8** @s, align 8
call void @_lfortran_strcpy(i8** %__main__global_stmts_2__1, i8* %0, i8 0)
br label %return
return: ; preds = %.entry
%1 = load i8*, i8** %__main__global_stmts_2__1, align 8
ret i8* %1
}
declare void @_lfortran_strcpy(i8**, i8*, i8)are you able to reproduce this same bug in normal/regular
lpythoninvocation?
No, this does not cause any problem with normal lpython execution.
Maybe we can optimize it to not make a copy. I will look into it.
src/lpython/python_evaluator.cpp
Outdated
| result.type = EvalResult::string; | ||
| result.str = r; | ||
| } else { | ||
| goto raise_error; |
There was a problem hiding this comment.
We don't use goto. Rather we can simply have throw LCompilersException("FortranEvaluator::evaluate(): Return type not supported"); here. (It might look like duplication, but I think its fine at the moment).
No description provided.