On Wed, 15 Jan 2025 at 20:57, Jakub Jelinek <ja...@redhat.com> wrote:
>
> On Wed, Jan 15, 2025 at 09:36:51PM +0100, Georg-Johann Lay via Gcc wrote:
> > > This pedwarn is correct, so I'm not sure why it's a problem. If you
> > > don't want warnings about non-standard extensions, don't use
> > > -pedantic-errors.
> >
> > The point is that I don't pedwarn explicitly, that's how the
> > testsuite works. What I am seeing is that with INT_N, many
> > tests in, say, avr.exp are failing which didn't fail with the
> > old definition of __int24. Similar will happen in user land,
> > which I'd like to avoid.
>
> __int128 a;
> emits such warning as well, there is nothing wrong with that and it is
> better when it is consistent.
> The tests can just stop using -pedantic-errors when they use the type
> (in some test subdirectories that is the default, so one needs to
> e.g. specify { dg-options "" } or some actual options to override it),
> or use
> __extension__ __int128 a;
> (then the pedantic warning and error are not emitted).
It looks like there are only 24 uses of __int24 in the testsuite, so
maybe just adding __extension__ to those places is the best solution.
> The reason for the pedwarn is that these are extensions to the standard,
> and the purpose of -pedantic or -pedantic-errors is diagnose extensions
> used in the program (and __extension__ a way to override that pedanticity).
> We pedwarn even about long long with -std=c89...
>
> Jakub
>
