https://bugzilla.redhat.com/show_bug.cgi?id=2112474
--- Comment #16 from John Snow <jsnow(a)redhat.com> ---
(In reply to Maxwell G from comment #15)
(In reply to John Snow from comment #11)
> - Need man pages for CLI scripts
I'd recommend manpages for your users, but this is not a requirement. It's a
SHOULD guideline, not a MUST.
The docs are there, I just need need to cajole Sphinx into writing a manpage
for it. It'd be a shame to not have them accessible in a manner that users
expect, I took the time to write them. Plus, I have a little bit of time while
I wait for others to help unblock me on the dependency problem I mentioned, so
I might have this for my next iteration.
> - Might need gpgverify (?)
You only need to use %gpgverify if upstream already provides signed
tarballs. You can ignore this.
I'm the upstream -- There's only one non-preview release, and I did indeed sign
it. I've already added this step to my working version.
(As an aside; being the upstream is why I am here trying to write the specfile,
as a courtesy to the other QEMU downstream packagers who will wind up needing
this package as a builddep for QEMU's own %check phase. I am realizing I went a
little out of order and there are other criteria I need to fulfill to become a
packager and own this package, but I am working on that process. I am a little
waylaid because I am moving right now, so attention has been scattered and time
has been short. I'm working on it. Feel free to gently remind me if I seem to
have missed something obvious -- I probably did, in the chaos.)
> - Need to integrate tests into a check phase, but there are dep problems.
What dependencies are missing? You should run tests if it's
possible/practical. If not, you should at least use %pyproject_check_import.
This serves as a basic smoke test by trying to import all public modules.
avocado-framework v90.0 or better. We use avocado upstream (in QEMU proper) and
I was asked to try using it for my Python subproject instead of the more usual
pytest to help dogfood that project, as well as to not add
yet-another-testing-tool to the QEMU ecosystem.
I'll add the smoke test for now, but see below for more on avocado.
> One interesting piece is that the .tar.gz for the SDist on PyPI
actually
> includes my .gitlab-ci.d files, one of which is licensed as GPLv2.I didn't
> really intend to distribute these files in the SDist, so I'll have to look
> into
> that. It's harmless for now, but could cause problems when I go to drop the
> remaining legacy GPLv2 code. Thought I'd mention it.
For Fedora, the only license that really matters is the license of the
actual installed content. The License tag does not need to account for other
files in the source tarball that aren't installed.
Good to know, thanks.
(I think I still want to look into how to drop them from the SDist to begin
with in the future if I can, but that's something for my research list.)
> I currently rely on avocado-framework >= 90, but I suppose
> 92.x (the "LTS" release) isn't packaged for F36 yet.
Yes, you'll have to edit the metadata to relax this requirement.
I can't - I am using asyncio features that only showed up in v90, and the tests
would have to be rewritten substantially to target the older version. I
couldn't find a configuration that works well across both versions, it's a
behavior change I am relying on.
Cleber Rosa (upstream maintainer for avocado-framework) tells me that
fedora:latest carries a stream wherein avocado is kept at the bleeding edge,
but that a non-modular package cannot rely upon a modular one as a dep. He is
working on updating the normal package from Avocado LTS 82.0 to Avocado LTS
92.0, which will unblock me here.
Thanks for the feedback! My action items right now are:
- Help Cleber get the avocado package updated
- Publish manpages for the CLI tools that I bundle. Will likely be rectified
for upstream's v0.0.2.
- Work through
https://docs.fedoraproject.org/en-US/package-maintainers/Joining_the_Pack...
--
You are receiving this mail because:
You are always notified about changes to this product and component
You are on the CC list for the bug.
https://bugzilla.redhat.com/show_bug.cgi?id=2112474