Hi Paul, the fix looks fine to me. Ok for mainline.
- Andre On Sat, 7 Dec 2024 09:17:57 +0000 Paul Richard Thomas <paul.richard.tho...@gmail.com> wrote: > Hi All, > > I must apologise for reintroducing this regression again, after the second > application of the fix for PR102689. I must admit that I had totally > forgotten about it, even though it was the reason for withdrawing the patch > the first time, and the failure was sporadic on my system, so I missed it > on regression testing both times around. > > I cannot see what is wrong with the element length coming from an explicit > typespec in the allocate statement but suppressing its use fixed the > problem. I don't especially like increasing the number of arguments in > gfc_array_init_size/gfc_array_allocate but it does improve clarity. I have > taken the opportunity to effect a temporary fix to the other regression > that I caused, PR117901, with the change to class_transformational_1.f90. I > prepared a proper patch for this but, in preparing a dedicated testcase, I > uncovered other problems that it didn't fix :-( > > Regression tests OK on Fedora/x86_64 - OK to push? > > Paul -- Andre Vehreschild * Email: vehre ad gmx dot de
