Does not build on 32-bit, but imho its much more urgent to fix Fedora 40 than to spend a lot of time on that. We can always do that later. Either way something is broken, but given the choice I'd rather have 32-bit broken.

Fine with me too to temporarily break *-i386-minimal

But i386 is just some one of several portability bugs that I reported to https://github.com/linbox-team/fflas-ffpack/issues and that are still open

FWIW, I'm packaging i686 without trouble, but we are on gcc 13.2.0. We regularly build and check sagemath on a few architectures including i686 (currently 10.4.beta5 at void-linux/void-packages#49571)

I've just rebuilt and checked givaro / fflas-ffpack / linbox from scratch (on i686) to make sure they are still working, and they seem fine.

BTW, when we switched to gcc 13 I had to add linbox-team/givaro#218, maybe you want to add that one too? (see: void-linux/void-packages#47712 (comment))

That is the only patch I have you don't (the other differences are: I don't have givaro 226 and I don't have linbox 310-backport, those seem to be needed for gcc 14).

So anyone wants to set it to positive review?

Reducing from blocker to critical because blockers still take effect as CI Fixes until this is merged:

• CI: Handle the "p: CI Fix" label #37950

Testing at https://github.com/mkoeppe/sage/actions/runs/8995091991

See #37698 (comment) , for some reason sage cannot detect installed givaro.
Might be relevant

See #37698 (comment) , for some reason sage cannot detect installed givaro. Might be relevant

It happens for both givaro and fflas-ffpack. In the cited config.log, system packages givaro and fflas-ffpack are marked as suitable (it is not the case for linbox, since flint is used as a package); but in the final summary they are marked as not suitable (there are the only switches).

It happens for both givaro and fflas-ffpack. In the cited config.log, system packages givaro and fflas-ffpack are marked as suitable (it is not the case for linbox, since flint is used as a package); but in the final summary they are marked as not suitable (there are the only switches).

See #37348 (comment)

Testing at https://github.com/mkoeppe/sage/actions/runs/8995091991

It builds OK on all Linux platforms except i386.
Let's merge it as suggested and sort out the breakage later.

For me, it works, the only issue is the strange situation with the system packages

I get

...
[fflas_ffpack-2.5.0] [spkg-install] libtool: link: g++ -std=gnu++11 -std=gnu++11 -dynamiclib  -o .libs/libfflas_c.1.dylib  .libs/fflas_lvl1.o .libs/fflas_lvl2.o .libs/fflas_lvl3.o .libs/fflas_sparse.o   -L/Users/kwankyu/GitHub/sage-dev/local/lib ./.libs/libfflas.dylib -lopenblas  -O2 -march=native -g -O2 -Wl,-rpath -Wl,/Users/kwankyu/GitHub/sage-dev/local/lib -Wl,-rpath -Wl,/Users/kwankyu/GitHub/sage-dev/local/lib   -install_name  /Users/kwankyu/GitHub/sage-dev/local/lib/libfflas_c.1.dylib -compatibility_version 2 -current_version 2.0 -Wl,-single_module
[fflas_ffpack-2.5.0] [spkg-install] Undefined symbols for architecture x86_64:
[fflas_ffpack-2.5.0] [spkg-install]   "Givaro::Integer::operator%(unsigned long long) const", referenced from:
[fflas_ffpack-2.5.0] [spkg-install]       Givaro::Modular<double, double, void>::init(double&, Givaro::Integer const&) const in fflas_lvl1.o
[fflas_ffpack-2.5.0] [spkg-install]       Givaro::Modular<double, double, void>::init(double&, Givaro::Integer const&) const in fflas_lvl2.o
[fflas_ffpack-2.5.0] [spkg-install]       Givaro::Modular<double, double, void>::init(double&, Givaro::Integer const&) const in fflas_lvl3.o
[fflas_ffpack-2.5.0] [spkg-install] ld: symbol(s) not found for architecture x86_64
[fflas_ffpack-2.5.0] [spkg-install] clang: error: linker command failed with exit code 1 (use -v to see invocation)
[fflas_ffpack-2.5.0] [spkg-install] make[7]: *** [libfflas_c.la] Error 1
[fflas_ffpack-2.5.0] [spkg-install] make[7]: *** Waiting for unfinished jobs....
[fflas_ffpack-2.5.0] [spkg-install] make[6]: *** [all-recursive] Error 1
[fflas_ffpack-2.5.0] [spkg-install] make[5]: *** [all-recursive] Error 1
[fflas_ffpack-2.5.0] [spkg-install] make[4]: *** [all-recursive] Error 1
[fflas_ffpack-2.5.0] [spkg-install] make[3]: *** [all] Error 2
[fflas_ffpack-2.5.0] [spkg-install] ********************************************************************************
[fflas_ffpack-2.5.0] [spkg-install] Error building fflas_ffpack-2.5.0

on mac with 3.5 GHz 6-Core Intel Xeon E5.

fflas_ffpack-2.5.0.log

Any help?

There is a -lgivaro missing, possibly intel mac only? New ticket, please