Re: [PATCH 3/2] c++: INIT_EXPR trial constexpr folding

On Sun, 16 Aug 2026, Jason Merrill wrote:

> On 8/14/26 2:25 PM, Jason Merrill wrote:
> > On 8/6/26 5:11 PM, Patrick Palka wrote:
> > > Bootstrapped and regtested on x86_64-pc-linux-gnu, does this look OK
> > > for trunk only?
> > 
> > OK.
> This seems to have regressed
> 
> > FAIL: 23_containers/mdspan/layouts/padded_neg.cc  -std=gnu++26  (test for
> > errors, line 247)
> > FAIL: 23_containers/mdspan/layouts/padded_neg.cc  -std=gnu++26  (test for
> > errors, line 248)
> > FAIL: 23_containers/mdspan/layouts/padded_neg.cc  -std=gnu++26  (test for
> > errors, line 298)
> > FAIL: 23_containers/mdspan/layouts/padded_neg.cc  -std=gnu++26  (test for
> > errors, line 302)
> > FAIL: 23_containers/mdspan/submdspan/canonical_slices_neg.cc  -std=gnu++26
> > (test for errors, line 58)
> > FAIL: 23_containers/mdspan/submdspan/canonical_slices_neg.cc  -std=gnu++26
> > (test for errors, line 61)
> > FAIL: 23_containers/mdspan/submdspan/canonical_slices_neg.cc  -std=gnu++26
> > (test for errors, line 62)

Ah sorry, dunno how I missed these failures when regtesting the patch
series last week.  Interestingly it seems only diagnostics have changed,
not the overall answer.