Thanks for your mail. It is very big of you to reconsider so
carefully and publicly.
Paride Legovini writes ("Re: Bug#1074556: autopkgtest: Drop the schroot virt
server"):
> 2. I think we can consider having a "code owner" for a-v-schroot, i.e.
> someone caring about the virt server bugs. This is the maintenance that
> would go in src:autopkgtest-virt-extra.
I am of course volunteering for this task. Please feel free to put my
email address in appropriate places. (Realistically I don't think I
could sensibly try to subscribe to the whole src:autopkgtest package.)
> 4. Even if we eventually decide to split out a-v-schroot, we should
> consider alternatives to a whole new source package. Possibilities I can
> think of: a separate _binary_ package, or a contrib/ directory, similar
> to what git does. I think the src:autopkgtest-virt-extra split was
> rushed, and it is probably not the best solution.
I agree that it's not a good solution. I think that maintaining this
code within src:autopkgtest will be easier from many practical points
of view.
Binary package structure can be discussed, of course. I don't feel
that the current situation is a problem, although I'd like to see the
autopkgtest virt protocol more widely appreciated and used.
> To conclude, I think it is reasonable for me to propose reverting my MR
> (https://salsa.debian.org/ci-team/autopkgtest/-/merge_requests/410),
> plus adding documentation on the fact that the a-v-schroot is a not a
> recommended virt server.
I don't agree that it's not recommended.
To put it more clearly: schroot itself is not disrecommended. It may
not be suitable for every purpose, of course. One should use software
that meets one's needs.
How about we put a note in the documentation for
autopkgtest-virt-schroot that explains that it doesn't provide strong
security isolation ? I'd be happy to write an MR for that. I would
ask the schroot maintainers' opinion about it. It's possible that
schroot itself could benefit from improvementsf to documentation of
this aspect. Also, I have heard that schroot has grown some further
unshare features. That would also need to be considered.
All of that doesn't need to make it non-recommended. I use schroot
(and autopkgtest-virt-schroot) it with code in the testbed that I
trust, precisely because often I find the imperfect security isolation
convenient. I think this is a common use case for Debian maintainers.
Regards,
Ian.
--
Ian Jackson <ijack...@chiark.greenend.org.uk> These opinions are my own.
Pronouns: they/he. If I emailed you from @fyvzl.net or @evade.org.uk,
that is a private address which bypasses my fierce spamfilter.
