Re: [PATCH] PR122597 -fdefault-integer-8 is results in wrong v_list in user defined IO

Hi Jerry,

Am 22.12.25 um 04:00 schrieb Jerry D:
> The attached patch from Steve issues an error if a user trys to use DTIO
> with -fdefault-integer-8. I reduced the test case and set it up for the
> testsuite.  Some may think this is a bit harsh, but one would get silent
> wrong results otherwise.

I think it is fine to reject it.  If you believe the message is too
harsh, how about:

"Sorry: -fdefault-integer-8 is not supported with user-defined 
input/output at %C"

I leave the final decision at your discretion.

OK for mainline, and thanks for the patch!

Harald

> Regression tested on x86_64.
> 
> OK for mainline?
> 
> Regards,
> 
> Jerry
> 
> Author: Steve Kargl <[email protected]>
> Date:   Sun Dec 21 18:32:46 2025 -0800
> 
>      fortran [PR122957] DTIO incompatibility with -fdefault-interger-8
> 
>      The -fdefault-integer-8 option is optional to assist with legacy
>      fortran codes. It is not a Stndard requirement and is not
>      compatible with the newer user defined derived type I/O.
> 
>              PR fortran/122957
> 
>      gcc/fortran/ChangeLog:
> 
>              * interface.cc (gfc_match_generic_spec): Issue an error
>              so that users do not use -fdefault-integer-8 with DTIO.
> 
>      gcc/testsuite/ChangeLog:
> 
>              * gfortran.dg/pr122957.f90: New test.
>