Re: Build failures with the Octave 4.4 transition
* Sébastien Villemot <sebastien@debian.org> [2018-06-11 08:22]:
The Octave 4.4 transition is more complex than anticipated. Many packages fail
to build, see:
https://udd.debian.org/dmd/?team%2Bpkg-octave-team%40tracker.debian.org#todo
https://release.debian.org/transitions/html/auto-octave.html
These are the problematic cases left:
- octave-interval: I hope that Oliver Heimlich is taking care of it.
- octave-stk: Idem for Julien Bect.
- octave-sparsersb: The upstream author, Michele MArtone, is currently
looking at the problem.
- octave-tisean: This package FTBFS because one of the 162 unit
tests is failing badly on amd64. Hovwever, for some other architectures,
there are issues related to numeric precision in several unit tests. We
should perhaps exclude this package from the transition.
- octave-ltfat: FTBFS mysteriously on mips and mipsel. Should we
exclude mips and mipsel from the list of allowed architectures?
- octave-odepkg: Does not compile against Octave 4.4. I tried to
build the package from the sources taken directly from the upstream
Mercurial repository (https://bitbucket.org/odepkg/odepkg, indicated
version is 0.9.1) but it FTBFS miserably with failures in several unit
tests. I think we should exclude this package from the transition.
- plplot: Fixing Bug#901504 may solve the problem (thanks, Sébastien!).
- octave-communications: FTBFS due to the inclusion of the
non-existent file octave/config.h. There is no commit in the upstream
Mercurial repository at SourceForge for fixing the bug. Should we
exclude this package from the transition, too?
- libsbml: FTBFS for reasons unrelated to Octave, apparently.
- sdpa: Has build-dependency problems on armel, mips, and mipsel (I
could not find out which are those problems)
Rafael
Reply to: