On Fri, Sep 23, 2022 at 5:33 AM Aron Xu <[email protected]> wrote: > Package: siconos > Version: 4.4.0+dfsg-1 > Severity: important > > The new version FTBFS on s390x in test suite 23, relevant log entry: > > > Cholesky solve preserving matrix > > Cholesky solve kNM_test: ./numerics/src/tools/NumericsMatrix.c:3023: > NM_csc: > Assertion `A->matrix2->csc->m == A->size0 && "inconsistent size > > Full build log is > > https://buildd.debian.org/status/fetch.php?pkg=siconos&arch=s390x&ver=4.4.0%2Bdfsg-1&stamp=1663797112&raw=0 > > Regards, > Aron > > Thanks for this bug report. I was able to reproduce the problem under qemu, and it is unclear to me whether the problem is actually a missing step in dense-to-sparse conversion, or simply a problematic assert after. It seems to be an issue showing up only on this architecture, because the numerical results of the test are slightly different, but this is possibly revealing a logical error in how dense to sparse is working or how the results are checked. Therefore, I have reported it upstream [0] and will try to include a patch in a package update if/when we are able to determine what is going on here.
[0] https://github.com/siconos/siconos/issues/467 regards, Steve
