On Wednesday, April 4, 2018 at 5:28:25 AM UTC+2, John H Palmieri wrote: > > > > On Tuesday, April 3, 2018 at 2:09:17 PM UTC-7, François Bissey wrote: >> >> >> >> > On 4/04/2018, at 02:34, John H Palmieri <jhpalm...@gmail.com> wrote: >> > >> > >> > >> > On Monday, April 2, 2018 at 9:25:56 PM UTC-7, François Bissey wrote: >> > >> > >> > > On 2/04/2018, at 16:27, John H Palmieri <jhpalm...@gmail.com> wrote: >> > > >> > > >> > > >> > > On Sunday, April 1, 2018 at 1:16:57 PM UTC-7, François Bissey wrote: >> > > >> > > >> > > > On 2/04/2018, at 03:48, John H Palmieri <jhpalm...@gmail.com> >> wrote: >> > > > >> > > > >> > > > >> > > > On Sunday, April 1, 2018 at 2:06:37 AM UTC-7, Volker Braun wrote: >> > > > On Sunday, April 1, 2018 at 9:01:17 AM UTC+2, Dima Pasechnik wrote: >> > > > - when I do an incremental upgrade, I do not see these failures. >> > > > - when I do 'make distclean; make ptestlong', I see them (with >> > > > 8.2.beta8 or later; I haven't checked earlier versions). >> > > > >> > > > Is it the other way round? Surely it should work with "make >> distclean; make ptestlong"? >> > > > >> > > > >> > > > It is not the other way around: failures when building from >> scratch, no failures when upgrading. >> > > > >> > > >> > > I (painfully) checked on my old laptop and I get this stuff when >> > > building from scratch too. So we have some kind of new >> mis-compilation >> > > with clang. >> > > >> > > François >> > > >> > > You made a guess in email that it was Singular. Suggestions for how >> to test this? >> > > >> > >> > Based on the libraries apparently involved. But looking at the chain of >> > calls, i would be more inlined to suspect something in sage itself. >> Either >> > in the glue to singular or in the coercion framework. >> > But that’s speculative, I have no real ideas. Does anyone know if that >> breaks >> > 8.1 as well? If not, we may be able to bisect something. >> > >> > François >> > >> > I just tried with 8.1, and all tests pass. That builds with gcc, not >> clang, so it's a big difference. >> >> Forgot about that! So the earliest we can test is 8.2.beta5. >> >> François >> >> > 8.2.beta5 doesn't build without #24704 ("ppl build fails with > clang-5.0"), in case that provides any clues. WIth #24704, I get the same > doctest errors. >
Oh, clang 5.0 is tricky. I did have trouble with it on #22679 (porting to FreeBSD), this is by the way how the patch from #24704 arose. > > -- > John > > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.