Re: [Patch, libgfortran] Add const restrict to intrinsic functions

[Janne Blomqvist <jblomqvi at cc dot hut dot fi>]

On Sat, Nov 26, 2005 at 03:57:16PM +0100, Thomas Koenig wrote:
> On Sat, Nov 26, 2005 at 02:58:30PM +0200, Janne Blomqvist wrote:
> > Hi,
> > 
> > the attached patch adds const restrict to the arguments for the
> > intrinsic functions in the m4/ directory.
> 
> I believe this is incorrect for cases like eoshift1, which has:
> 
>   if (ret->data == NULL)
>     {
>       int i;
> 
>       ret->data = internal_malloc_size (size * size0 ((array_t *)array));
>       ret->offset = 0;
>       ret->dtype = array->dtype;
> 
> You can't declare ret as "const" in this case.

I think I can, since when the "const" is after the "*" it applies to
the pointer itself, not whatever the pointer points to.

And yes, it doesn't really matter outside the function, since C
argument passing is passing a copy of the pointer. So in this case
"const" is just a reminder for the libgfortran programmer not to go
and mess with the pointer since that will almost certainly mess stuff
up.

-- 
Janne Blomqvist

Attachment: pgp00000.pgp
Description: PGP signature