FTBFS with GCC-14 (rebuild against mpi-defaults 1.17)

[Erich Eickmeyer]

Following a no-change rebuild against mpi-defaults 1.17, these packages completely FTBFS on all architectures.

This appears to be due to GCC-14.

See individual package builds for more details.

[Erich Eickmeyer (eeickmeyer)]

Strangely, fftw does *not* fail on armhf, though it does fail on all other archs.

[Erich Eickmeyer (eeickmeyer)]

Similarly to fftw, hypre also ftbfs !armhf

[Erich Eickmeyer (eeickmeyer)]

Re-added ruby-mpi since there is, indeed, some weird gcc madness going on there.

[Erich Eickmeyer (eeickmeyer)]

This bug was fixed in the package hypre - 2.31.0-1exp2

---------------
hypre (2.31.0-1exp2) experimental; urgency=medium

  * debian patch build_gcc14_PR1110.patch applies upstream PR#1110 to
    enable build using gcc-14. Closes: #1075077.

 -- Drew Parsons <email address hidden> Sun, 04 Aug 2024 11:17:02 +0200

hypre (2.31.0-1exp1) experimental; urgency=medium

  * New upstream release. New ABI 2.31.0.
    - applies debian patches fix_make_check.patch,
      no_timings_finalizeAllTimings.patch
    - obsoletes debian patch test.patch (scripts now use bash)
  * add Drew Parsons as uploader

 -- Drew Parsons <email address hidden> Sat, 20 Jul 2024 14:00:52 +0200

[Erich Eickmeyer (eeickmeyer)]

For scalapack: the solution? Use GCC-13 or older per https://github.com/Reference-ScaLAPACK/scalapack/issues/97. (obviously not an option)

[Launchpad Janitor (janitor)]

This bug was fixed in the package sundials - 6.4.1+dfsg1-3ubuntu1

---------------
sundials (6.4.1+dfsg1-3ubuntu1) oracular; urgency=medium

  * Use PetscInt instead of sunindextype where PETSc functions
    are called to fix FTBFS with GCC 14 (LP: #2075261)
  * Cherry-pick upstream commit to use int64_t for
    KLU_INDEXTYPE to fix FTBFS on armhf

 -- Graham Inggs <email address hidden> Sun, 22 Sep 2024 11:18:32 +0000

[Graham Inggs (ginggs)]

Fixed in sopt 4.1.0+dfsg-1