Paul Richard Thomas wrote: > This PR is caused by a rather unfortunate cast for the pointer to the > return value from matmul: [...] > The fix is obvious (ie use its natural kind) [...] > Dominique points out that the patch fixes PR58883 as well. I'll > enhance the testcase appropriately.
Nice! > Unless anybody has any objections, I'll commit to trunk tonight and to > 4.8 and 4.9 next weekend. Looks good to me. As the patch shouldn't have side effects and as Jakub intents to do a 4.9.1 RC early this week, I'd prefer it the 4.9 patch would be committed earlier. Tobias
