[libgfortran,patch] Zero-length strings messed up in library (PR33079)

[Tobias Schlüter]

FX Coudert wrote:
>>> -  if (*len > 0)
>>> +  if (*len == 0)
>>> +    {
>>> +      /* A zero-length Fortran string is "".  */
>>> +      char * tmp = internal_malloc_size (1);
>>> +      tmp[0] = '\0';
>>> +      *dest = tmp;
>>> +    }
>>> +  else
>>
>> Do you zero-terminate the string because you don't want to allocate 
>> zero memory?  I don't know if allocating a pointer to zero memory 
>> would work, though.
> 
> I don't think you can portably call malloc(0) and expect a non-NULL 
> pointer out of it. From 
> http://www.opengroup.org/onlinepubs/009695399/functions/malloc.html :
>> If the size of the space requested is 0, the behavior is 
>> implementation-defined: the value returned shall be either a null 
>> pointer or a unique pointer.
> Also, internal_malloc_size catches this case and returns NULL, so I 
> think it's more reasonable to special-case zero right there. (And I do 
> set the allocated memory to zero just to make sure we don't end up with 
> unassigned memory.)

An alternative would be to have a
  static char zero_length_string[0]; // Maybe [1]
at file scope and then do
  *dest = zero_length_string;
This would evade the unnecessary allocations.

Your patch is ok, if you don't like my suggestion better.

- Tobi