Hi Simon,

Thanks for the reply.

Honest answer: I do not have a strong attachment to test_the_archive
itself. What bit me was operational: after Flight Deck import, every
push showed a red/stuck pipeline (tag go-ci, no runner, no job trace).
As a new uploader without project Settings, that looked like "CI is
broken / I did something wrong", and the docs still present go-ci +
test_the_archive as the default path via debian/gitlab-ci.yml.

For catching packaging mistakes I rely more on local debuild/lintian
and (when it runs) Salsa CI. If the team consensus is that
test_the_archive is low value and the future is debian/salsa-ci.yml +
shared runners, I am happy to follow that and not spend effort on
another go-ci box.

The runner offer still stands if someone wants spare go-ci capacity;
otherwise I will treat weekend go-ci outages as non-blocking and focus
on Salsa CI / local smoke.

Thanks,
Hermes Rodriguez <[email protected]>

El lun, 14 sept 2026 a las 3:59, Simon Josefsson
(<[email protected]>) escribió:
>
> Hermes Jesus Rodriguez Azuaje <[email protected]> writes:
>
> > Hi,
> >
> > I have been packaging the ncruces / pgwd chain under go-team/packages.
> > Over the weekend several test_the_archive jobs stayed stuck / failed with
> > no matching runners for tag go-ci (no job trace). That matches a single
> > point of failure on the current go-ci pool.
>
> What value do you find in the test_the_archive CI job?  I'm not
> objecting to fixing it, but I never found it useful for any of my
> workflows.  When does it help you notice a problem in the packaging?
> For me it normally just always succeeds regardless of package mistakes,
> and sometimes it consistently fail (e.g., for packages with repacked
> upstream sources) even though there is no problem with the package.
>
> /Simon

Reply via email to