How come same version pep8 does't work the same?

[Dan Kenigsberg]

On Thu, May 22, 2014 at 05:45:45PM +0200, David Caro wrote:
> On Thu 22 May 2014 03:50:23 PM CEST, Yaniv Dary wrote:
> >
> >
> > ----- Original Message -----
> >> From: "Vered Volansky" <vered at redhat.com>
> >> To: "infra" <infra at ovirt.org>
> >> Sent: Thursday, May 22, 2014 4:48:37 PM
> >> Subject: How come same version pep8 does't work the same?
> >>
> >> I see a patch failing to build vsdm rpm on a pep8 error. pep8 version is
> >> 1.4.6, yet the actual pep8 job works jst fine. Same pep8 version to both.
> >> How come?
> >
> > Please upgrade your version of pep8.
> > There were many changes in the latest update and this is probably causing your issue.
> >
> >
> > Yaniv
> >
> >> I need to know why this happens in order to fix my job.
> >>
> >> Patch in question: http://gerrit.ovirt.org/#/c/26759/5
> >> pep8 job: http://jenkins.ovirt.org/job/vdsm_master_pep8_gerrit/9179/console
> >> Failing job:
> >> http://jenkins.ovirt.org/job/vdsm_master_storage_functional_tests_localfs_gerrit/852/consoleFull
> >>
> >> Thanks,
> >> Vered
> >> _______________________________________________
> >> Infra mailing list
> >> Infra at ovirt.org
> >> http://lists.ovirt.org/mailman/listinfo/infra
> >>
> > _______________________________________________
> > Infra mailing list
> > Infra at ovirt.org
> > http://lists.ovirt.org/mailman/listinfo/infra
> 
> Couple of things to have in mind here:
> 
> The jobs that you sent are for different patches, I'll focus on the 
> patch you sent only.
> 
> Why pep8 job did pass:
>   - Because the pep8 job only checks the changes made in that patch, 
> and in that patch, there were no issues (git diff HEAD~)
> 
> Why vdsm_master_storage_functional_tests_localfs_gerrit did not pass:
>   - Because when compiling it will check all the files, not only the 
> ones your patch changes, and your patch was based on an already failing 
> patchset (http://gerrit.ovirt.org/#/c/27977/5/vdsm/BindingXMLRPC.py,cm).
> 
> You can see in the gerrit comments for that patchset that it did 
> actually fail:
> Patch Set 5: Code-Review-1 Verified-1
> Build Unstable
> http://jenkins.ovirt.org/job/vdsm_master_pep8_gerrit/9162/ : UNSTABLE
> http://jenkins.ovirt.org/job/vdsm_master_unit_tests_gerrit/9305/ : 
> SUCCESS
> http://jenkins.ovirt.org/job/vdsm_master_unit_tests_gerrit_el/8374/ : 
> SUCCESS
> 
> 
> Mybe it's a good point to run pep8 on all the files and not only on the 
> latest commit diff, to block any patchset based on a previously failing 
> patch. Ideas?

I prefer to keep the diff. The previously-failing patch should be marked
and fixed. If follow-ups are fine - let them be.
Skipping pep8 (with PEP8=true) during the storage job would have avoided
the issue in a cleaner way.