On Thu, Jul 10, 2025 at 12:05 PM Lauve, Aaron <ala...@luc.edu> wrote: > > Thanks, Dima. > > > > “Yes” to using gfortran provided by homebrew (at least I presume) > > Here’s what brew has to say about that: > > > > lauve@edinburgh sage % brew info gfortran > > ==> gcc: stable 15.1.0, HEAD > > GNU compiler collection > > https://gcc.gnu.org/ > > Installed > > /opt/homebrew/Cellar/gcc/15.1.0 (1,957 files, 492.3MB) * > > Poured from bottle using the formulae.brew.sh API on 2025-06-24 at 09:16:18 > > From: https://github.com/Homebrew/homebrew-core/blob/HEAD/Formula/g/gcc.rb > > License: GPL-3.0-or-later WITH GCC-exception-3.1 > > ==> Dependencies > > Build: make ✘ > > Required: gmp ✔, isl ✔, libmpc ✔, mpfr ✔, zstd ✔ > > ==> Options > > --HEAD > > Install HEAD version > > > > I notice that ‘gcc’ mentioned above is 15.1.0 while the relevant line in the > config file is 14.2.0. 14.2.0 is Sage's version of gfortran, but there are lines
real_configure:11002: checking for gfortran real_configure:11023: found /opt/homebrew/bin/gfortran real_configure:11035: result: gfortran real_configure:11061: checking for Fortran compiler version real_configure:11070: gfortran --version >&5 GNU Fortran (Homebrew GCC 15.1.0) 15.1.0 [...] real_configure:18788: will use system package and not install SPKG gfortran which really say what it is, and what's going to be used. > > Attaching the toplevel config.log file. right, I knew about this, I mentioned to Homebrew people that it's a bug: https://github.com/Homebrew/homebrew-core/issues/209091 but they decided not to do anything about it. The problem is that `openblas.pc` supplied by them does not work with Apple's "gcc", i.e. clang. It only works with "real" gcc, which we'd rather avoid using in order not to create even more mess than what's our current build system already is. A quick an dirty fix is to replace their `openblas.pc` with the corrected one, where openmp is not mentioned at all - in the same vein as done in https://github.com/sagemath/sage/pull/40397 (I'll post such a file somewhere) As OpenBLAS people were more keen to listen at https://github.com/Homebrew/homebrew-core/issues/209091 they https://github.com/OpenMathLib/OpenBLAS/pull/5180 and so it won't be hard to create a custom Homebrew tap with the alternative, working with clang, configuration. Although I think Accelerate is not the better path to go forward. > > > > > > Also, maybe “yes” to being an Apple Accelerate Guinea pig. Let’s see what > transpires… Thanks! I've just made a draft PR from the branch I mentioned https://github.com/sagemath/sage/pull/40397 Would most appreciate feedback from you and any other volunteers. Cheers, Dima > > > > > > > > -- > > Aaron Lauve (he/him/his) > > Drop in for a visit > > 773.508.3727 | BVM 501 | ala...@luc.edu > > ~~~~~~~~~~ > > Associate Professor and Graduate Program Director for Mathematics > > Department of Mathematics & Statistics > > Loyola University Chicago > > > > From: Dima Pasechnik <dimp...@gmail.com> > Date: Thursday, July 10, 2025 at 11:54 AM > To: sage-devel@googlegroups.com <sage-devel@googlegroups.com>, Lauve, Aaron > <ala...@luc.edu> > Subject: Re: [sage-devel] Re: openblas fail on build from source > > [You don't often get email from dimp...@gmail.com. Learn why this is > important at https://aka.ms/LearnAboutSenderIdentification ] > > Hi Aaron, > thanks for the report. > The email interface to google groups allows bigger attachments, by the way. > (you can also email these logs directly to me) > > There could be several reasons for all this, but first of all, it's > good to understand why the Homebrew's openblas is not accepted. > Can you send us the top-level config.log ? > > As far as building from source is concerned, I assume you use gfortran > from Homebrew, right? > It could be that 1.3.28 is too old for your OS/hardware, and we need > to bump our package version. > > Last but not the least, I'm at the moment working on allowing the user > to rather use the Apple Accelerate framework instead of openblas, see > https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fdimpase%2Fsage%2Ftree%2Faccel_wip&data=05%7C02%7Calauve%40luc.edu%7C66bdbab0def2439a0b0b08ddbfd27cfa%7C021f4fe32b9c48248378bbcf9ec5accb%7C0%7C0%7C638877632994015794%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=hfdKe%2FfqH%2BrfN%2Fbi7gZXn1xaFoZm%2FcshCLKopIqAegQ%3D&reserved=0 > If you're interested in testing the latter, I can provide more instructions. > > Best > Dima > https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpasechnik.info%2Fdima%2F&data=05%7C02%7Calauve%40luc.edu%7C66bdbab0def2439a0b0b08ddbfd27cfa%7C021f4fe32b9c48248378bbcf9ec5accb%7C0%7C0%7C638877632994054935%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=ppL6jxXxs2%2F2Ppv5K95jvyzWJxiSZ9UdjWAPHC%2BnC04%3D&reserved=0 > > > > > > On Thu, Jul 10, 2025 at 11:16 AM Aaron Lauve <la...@math.luc.edu> wrote: > > > > oops. forgot to attach the log. and then GoogleGroups complained it was too > > long. > > I'm attaching 'head -n1000' and 'tail -n1000' instead of the full log. Hope > > that's helpful. > > > > -- > > 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 view this discussion visit > > https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgroups.google.com%2Fd%2Fmsgid%2Fsage-devel%2Ff5c7a22b-2dee-425d-bd0b-e540cffbce5fn%2540googlegroups.com&data=05%7C02%7Calauve%40luc.edu%7C66bdbab0def2439a0b0b08ddbfd27cfa%7C021f4fe32b9c48248378bbcf9ec5accb%7C0%7C0%7C638877632994081750%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=Iwm2GGTj%2FyADuVPv8aPns0uF5bnTXdHMF6kttW1lpok%3D&reserved=0. -- 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 view this discussion visit https://groups.google.com/d/msgid/sage-devel/CAAWYfq281qFw7iZShYh8gEpt1mfzXfmzUFNnV3bV07M1q4QjqQ%40mail.gmail.com.
