[Bug fortran/93253] Regression on non-standard hex constant syntax

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93253

kargl at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |kargl at gcc dot gnu.org

--- Comment #1 from kargl at gcc dot gnu.org ---
(In reply to Dave Love from comment #0)
> I tried to build the NAS benchmarks with snapshot 'GNU Fortran (GCC) 10.0.0
> 20200105 (experimental)' and got:
> 
>   cd ../common; gfortran-10 -c  -Ofast -march=native -std=legacy -fdec -g
> randi8.f
>   randi8.f:26:20:
>   
>      26 |       data i246m1/X'00003FFFFFFFFFFF'/
>         |                    1
>   Error: Hexadecimal constant at (1) uses nonstandard syntax
>   randi8.f:57:20:
>   
>      57 |       data i246m1/X'00003FFFFFFFFFFF'/
>         |                    1
>   Error: Hexadecimal constant at (1) uses nonstandard syntax
> 
> Gfortran 8 doesn't complain, even without the "-std=legacy -fdec".  (I
> don't know where this syntax originates, but I guess DEC.)
> I haven't tried v9.
> 
> [I've also reported an NPB bug, but I guess NPB isn't alone.]

Do you read the document that comes with your compiler?

Try adding the -fallow-invalid-boz option to your command line.