Re: [patch, commit] Bug 104827 - [13/14/15/16 Regression] ICE in gfc_conv_array_constructor_expr,

Paul Richard Thomas <[email protected]> Tue, 7 Apr 2026 22:33:23 +0100

Newsgroups	gmane.comp.gcc.patches,gmane.comp.gcc.fortran
Message-ID	<CAGkQGiL_iiPixNa2yFM+RTSnfVkh1ifTkB250c0o5tVS-Lup3A@mail.gmail.com>

Hi Jerry,

That's fine. Passes that are untested do not need backposting.

Cheers

Paul

On Tue, 7 Apr 2026 at 17:37, Jerry D <[email protected]> wrote:
>
> I plane to commit the new test case provided by Christopher in the PR104827 and
> close it. The PR was likely fixed earlier in 15.
>
> Regards,
>
> Jerry
>
> commit 5f22ec5e26288e661ef6c2b93fd3948ad0aee429 (HEAD -> master)
> Author: Christopher Albert <[email protected]>
> Date:   Wed Mar 11 00:26:14 2026 +0100
>
>      fortran: Add testcase [PR104827]
>
>      The reproducer from the bug report now gives a proper error on current
>      trunk instead of an ICE.  Add a testcase to prevent future regressions.
>
>              PR fortran/104827
>
>      gcc/testsuite/ChangeLog:
>
>              PR fortran/104827
>              * gfortran.dg/gomp/pr104827.f90: New test.
>
>      Signed-off-by: Christopher Albert <[email protected]>
>
>
>
>