[Piuparts-devel] Piuparts cruft patch

[David Steele]

On Fri, Mar 15, 2019 at 2:07 PM Ivo De Decker <ivodd at debian.org> wrote:
>
> Hi David,
>
> On 3/14/19 1:10 PM, David Steele wrote:
> > Hi Ivo,
> >
> > I'm writing because there is an issue with the summary.json reports
> > coming from Piuparts. I wrote that section of the service, before
> > stepping away from the project.
> >
> > Last October, you authored a patch for piuparts that avoids reporting
> > failures in summary.json for package tests covering obsolete versions.
> >
> > It looks like there is a bug in the patch - upgrade sections incorrectly
> > show a majority of packages in the "waiting" state. You can see this now
> > on the piuparts column on your DDPO page.
> >
> > My question is, why did you submit that patch? Was there a real-world
> > problem with the temporary failure notification, or just an inconvenience?
>
> The patch was inspired by an issue with a package (I don't remember
> which one) that was blocked from migrating to testing because an old
> binary was left in unstable, which has a piuparts error. The newer
> version of that source package didn't have the binary. The patch was
> meant to filter out the result for these outdated binaries. The fact
> that this affects other packages was clearly not intended.
>
> > I ask because there appears to be little chance that this will be fixed
> > any time soon. Pejacevik is lacking of attention lately, and I don't see
> > a likely path for fixing the bug. Is there a compelling reason not to
> > revert the commit?
>
> The situation that my patch should have fixed is relatively infrequent,
> and there are 2 other ways to work around it:
>
> - Have the binary decrufted in unstable by the ftp-team (at that time,
> there was quite a backlog in handling the decrufts). This is only
> possible if the decruft doesn't break other packages in unstable.
>
> - An ignore-piuparts-hint by the release team to ignore the piuparts
> error (obviously after checking that it only affects the old binary).
>
> So if this patch is causing trouble, I'm not opposed to reverting it
> till a better fix is found (even if that could take a while).
>

Thanks for the reply.

It sounds like this is a situation that, when it crops up, should have
a bit of a spotlight shown (or an automated decrypt on ftp).

FYI - I intend to revert the patch.

Thanks again.

-- 
AE0D BF5A 92A5 ADE4 9481  BA6F 8A31 71EF 3661 50CE