On Sat, 26 Oct 2024 at 21:45:15 +, Daniel Markstedt wrote:
> The autopkgtest docs suggest that by putting a file in a particular
> directory would have it picked up as a test artifact for the CI job.
Yes. During your test, the name of that directory is given by the
environment variable AUTOPKG
Hi all,
The autopkgtest docs suggest that by putting a file in a particular directory
would have it picked up as a test artifact for the CI job.
Since Debian CI does provide a tarball with artifacts for each job, I was
hoping that you could put an arbitrary artifact from a test run and have the
Hi John (2024.10.26_08:45:15_+)
> > tox tests in Debian package builds are a little different because we use
> > --system-site-packages virtualenvs, but they can be a good way to deal
> > with this.
>
> Can you name any package using this mechanism so I can have a look?
wheel uses tox tests,
On Sat, 26 Oct 2024 at 11:02:29 +0200, Joachim Zobel wrote:
> I have tried all available --boot options for autopkgtest with s390x
> and was not able to create an image
I suspect you mean autopkgtest-{build,virt}-qemu. autopkgtest itself (the
test runner) does not have a --boot option.
The ci.deb
On Fri, Oct 25, 2024 at 03:03:53PM +, Holger Levsen wrote:
> the current expectation is that an NMU bug is opened, which contains
> the debdiff.
>
> https://www.debian.org/doc/manuals/developers-reference/developers-reference.en.html#when-and-how-to-do-an-nmu
>
> "... Then, you must send a pa
On Sat, 2024-10-26 at 11:08 +0200, John Paul Adrian Glaubitz wrote:
> I just realized that this doesn't work because kiwi doesn't have a setup.py
> anymore but just uses pyproject.toml. Do you know how it works in this case?
OK, I just remove every custom override and let debhelper and pybuild jus
Hi Timo,
On Sat, 2024-10-26 at 10:15 +0200, John Paul Adrian Glaubitz wrote:
> On Fri, 2024-10-25 at 22:19 +0200, Timo Röhling wrote:
> > I ran into this issue with old-style setuptools packages (i.e.,
> > packages with a setup.py); AFAIK the entry_points mechanism needs
> > valid egg-info or di
Am Freitag, dem 25.10.2024 um 08:08 +0200 schrieb Paul Gevers:
> > 2. I am unable to reproduce the failing test since there is no working
> > --boot option for autopkgtest and and I can't find any info on how to
> > run qemu for that architecture.
>
>
> I don't understand the remark in the first
Hello Stefano,
On Sat, 2024-10-26 at 02:45 +, Stefano Rivera wrote:
> If it actually needs the .dist-info to be on path (fully installed), you
> can try runnig the tests under tox. That's probably what the upstream
> does.
>
> tox tests in Debian package builds are a little different because
Hi Timo,
On Fri, 2024-10-25 at 22:19 +0200, Timo Röhling wrote:
> I ran into this issue with old-style setuptools packages (i.e.,
> packages with a setup.py); AFAIK the entry_points mechanism needs
> valid egg-info or dist-info metadata in the PYTHONPATH. My
> workaround was
>
>
> export PYBU
10 matches
Mail list logo
