sci-libs/med: fix build against hdf5-1.12 #22096
Conversation
Package depends on sci-libs/hdf5[mpi] which currently has test issues and has USE=mpi masked. Signed-off-by: Bernd Waibel <waebbl-gentoo@posteo.net>
Thanks to Alexandre Ferreira for providing the patch. Closes: https://bugs.gentoo.org/809008 Package-Manager: Portage-3.0.20, Repoman-3.0.3 Signed-off-by: Bernd Waibel <waebbl-gentoo@posteo.net>
Pull Request assignmentSubmitter: @waebbl sci-libs/med: @tamiko, @waebbl, @gentoo/sci, @gentoo/proxy-maint Linked bugsBugs linked: 809008 In order to force reassignment and/or bug reference scan, please append Docs: Code of Conduct ● Copyright policy (expl.) ● Devmanual ● GitHub PRs ● Proxy-maint guide |
gentoo-bot
added
self-maintained
Pull request CI reportReport generated at: 2021-08-24 21:14 UTC There are existing issues already. Please look into the report to make sure none of them affect the packages in question: |
|
Thanks 👍 |
|
@AndrewAmmerlaan Andrew, can you revert this? I just got an email that the mask breaks gmsh. Haven't tested the revdeps on this. |
Reverted, I'm slightly confused though, because the qa-bot said it was fine |
Does it check for revdeps? Just looked up latest gmsh and noticed it depends on med[mpi] unconditionally and has a Need to figure out how to re-track the branch again. Already deleted it locally :/ |
It should, it checks the whole tree. Possibly your local tree was significantly behind the main tree? |
No, I don't think so. I always rebase before pushing, so the local tree was up to date. |
|
Just checked the gmsh-4.8.4 build file. From my understanding the packages doesn't need to have mpi support for med enabled unconditionally. So instead of masking gmsh[med] as well, I suggest to change the dependency on sci-libs/med of all gmsh versions starting 4.6.0 and above from |
|
@AndrewAmmerlaan I noticed, that med will silently downgrade hdf5 to v1.10 whenever USE=mpi is passed to the ebuild. So we don't even need the mask and everything is fine. What do you think? Is this ok, the way it is? |
The issue with mpi is only present in >=1.12? In that case, can we not just mask the mpi flag for only 1.12+? |
The mask is already for >=hdf5-.1.12.1 only. |
Sorry, I'm a bit confused, if the mask is only on 1.12 then why do we need to mask mpi on med in the first place? We could just use the older version of hdf5 with mpi enabled to satisfy the dependency right? |
Yeah that's what I wanted to say in this comment.
Guess I didn't realize this in the first place. |
Aaah right, now I get it :P I think it is fine as is in that case |
Thanks to Alexandre Ferreira for providing the patch.
Closes: https://bugs.gentoo.org/809008
Package-Manager: Portage-3.0.20, Repoman-3.0.3
Signed-off-by: Bernd Waibel waebbl-gentoo@posteo.net
profiles/base/package.use.mask: mask mpi for sci-libs/med
Package depends on sci-libs/hdf5[mpi] which currently has test issues
and has USE=mpi masked.