Re: [Pkg-octave-devel] Octave 4.2 for stretch
* Mike Miller <mtmiller@debian.org> [2016-06-29 09:33]:
In case you are not aware,
1. Stretch will be defaulting to GCC 6 [1]
2. Octave 4.0.x FTBFS with GCC 6 [2][3]
3. Octave 4.2.0 is coming, hopefully in time for the stretch freeze [4]
The changes needed to make it build with GCC 6 were quite disruptive. I
don't think anyone has yet tried or proposed trying to patch 4.0.x to
make it work with GCC 6, but it looks like trouble.
Thanks for this report, Mike.
Any thoughts on schedule, mitigation plans, early coordination we could
do to have a successful Octave release on stretch? When would be a good
time to have a beta release in experimental considering the 2016-11-05
transition freeze?
We could release a 4.1.0+ hg snapshot version immediately to
experimental. I tried to compile Octave inside an experimental chroot
with sources checked out from the default branch. It fails with this
error message:
libinterp/octave-value/ov-cell.cc: In member function 'virtual bool octave_cell::load_hdf5(octave_hdf5_id, const char*)':
libinterp/octave-value/ov-cell.cc:1156:57: error: invalid conversion from 'octave_hdf5_err (*)(octave_hdf5_id, const char*, void*) {aka int (*)(int, const char*, void*)}' to 'H5G_iterate_t {aka int (*)(long int, const char*, void*)}' [-fpermissive]
hdf5_read_next_data, &dsub);
^
In file included from /usr/include/hdf5/serial/hdf5.h:31:0,
from libinterp/corefcn/oct-hdf5.h:31,
from libinterp/octave-value/ov-cell.cc:45:
/usr/include/hdf5/serial/H5Gpublic.h:165:15: note: initializing argument 4 of 'herr_t H5Giterate(hid_t, const char*, int*, H5G_iterate_t, void*)'
H5_DLL herr_t H5Giterate(hid_t loc_id, const char *name, int *idx,
^~~~~~~~~~
The libhdf5-dev package has version 1.10.0-patch1+docs-1~exp in
experimental while, in unstable, it has version 1.8.16+docs-8. Something
changed between these versions that make Octave FTBFS.
Do you know if there is any fix for it ?
Rafael
Reply to: