Anyone seen this one?

Jerry DeLisle jvdelisle@verizon.net
Tue Nov 29 05:16:00 GMT 2005


Steve Kargl wrote:
> On Mon, Nov 28, 2005 at 08:15:55PM -0800, Jerry DeLisle wrote:
> 
>>Andrew Pinski wrote:
>>
>>>>Hi all,
>>>>
>>>>Ran into this while testing some other stuff.  This is cute!  Compiles 
>>>>fine with the -std=f95 removed.
>>>>
>>>>$ gfc -std=f95 read_comma.f
>>>
>>>
>>>This is because abort is a non standard intrincs.  there is a PR About 
>>>this but it is
>>>unconfirmed because nobody knows the correct thing to do.
>>>
>>>
>>
>>Well I just put a bug report in not knowing any better.  I guess thats OK 
>>since nobody knows what to do.  :)
>>
> 
> 
> Try this patch of intrinsics/abort.c.  This is untested.  If it
> works for you and your regression test it.  Please commit.
> 
> 
> Index: abort.c
> ===================================================================
> --- abort.c	(revision 107569)
> +++ abort.c	(working copy)
> @@ -38,3 +38,15 @@ void PREFIX(abort) (void)
>    close_units ();
>    abort ();
>  }
> +
> +/* abort() is needed for the testsuite when linking with -std=f95.  */
> +
> +extern void abort_ (void);
> +export_proto_np(abort_);
> +
> +void
> +abort_ (void)
> +{
> +  close_units ();
> +  abort ();
> +}
OK I will give it a try.

Jerry



More information about the Fortran mailing list