Is it time to take the middle-end stringop/array warnings out of -Wall?

[Eric Gallager]

On Thu, May 21, 2026 at 12:08 PM Sam James via Gcc <gcc@gcc.gnu.org> wrote:
>
> Arsen Arsenović <aarsenovic@baylibre.com> writes:
>
> > Joseph Myers <josmyers@redhat.com> writes:
> >
> >> The point is that if using -Wextra, you should expect some warnings
> >> that are "hard to avoid ... no simple way to modify the code to
> >> suppress the warning" (whereas such warnings are outside the
> >> documented scope of -Wall).
> >
> > Yes, I agree.
> >
> > I'm just saying that I think that, beyond just moving into -Wextra, the
> > wording needs to be made less certain ("there is" -> "there may be" or
> > so).
>
> So, something like:
> 1) move to -Wextra;
> 2) change phrasing;
> 3) add diagnostic group -Wmiddle-end (name pending);
> 4) drop from -Wextra now you can conveniently use them via the group.
>
> then?

I agree with doing steps 1 through 3 listed here; I think that if 4 is
to be done, it should wait a release until people have had time to
adjust to the new -Wmiddle-end existing, though.

>
> sam