Details
- Reviewers
gerald tcberner - Commits
- R11:690ee1ef6bb0: math/armadillo: Update to 11.1.1
Diff Detail
- Repository
- R11 FreeBSD ports repository
- Lint
Lint Not Applicable - Unit
Tests Not Applicable
Event Timeline
I could not find a changelog for this minor version, see http://arma.sourceforge.net/docs.html#changelog. That's why my commit message is only the summary.
As for testing, I could not test any i386 version because -- again -- sysutils/slurm-wlm is in the dependency tree, which is NOT_FOR_ARCHS=i386. Since it is the third time that I find this issue, I have started to work to resolve it, see https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=262010#c9 . In case all attempt to fix it fail, I will consider setting a NOT_FOR_ARCHS variable for this port too.
All tests are successfull on amd64.
And I think spending much time on ports related to HPC and scientific computing specifically to get them to work on i386 is mostly wasted time.
That is, we should try not to break anything, but if it is broken already ... mark it thusly (or NOT_FOR_ARCHS) and don't worry too much.
Thanks Gerald for your advice. After having read it, I did some research and noticed on https://www.freebsd.org/platforms/ that with FreeBSD 13 the i386 architecture is downgraded from tier 1 to tier 2, so I see that you are indeed right.
However, in this particular case:
- the fix should not be too hard: I think adding an option to net/mpich is sufficient to fix most ports and someone was already working at updating it (again, see https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=262010 );
- according to https://www.freebsd.org/security/#sup , stable/12 is going EOL on June 30, 2024 and on 12 the i386 architecture is still tier 1.
So I think it worths a try. That said, if things get harder, I will surely follow your advice and just set NOT_FOR_ARCHS.