Responsibility for rebuilding dependent components, was: F-16 Branched report: 20110920 changes

Nicolas Chauvet kwizart at gmail.com
Tue Sep 20 22:56:31 UTC 2011


2011/9/20 Jesse Keating <jkeating at j2solutions.net>:
...
>>> thus we have bodhi
>>> and updates-testing as a gateway to get into the release.
>>
>> It's a gateway, I just don't think it serves as useful a purpose as it was intended to.
>
>
> The question though really is whether or not it is more useful than a few (like 4) release engineers looking at trac tickets and making best guesses depending on whatever the ticket reporter put in the ticket about the change.  We were trying to improve the workflow, not perfect it.  Has it been improved?

I think the situation improved with bodhi buildroot overrides over trac tickets.
But I've hit several issues with the opencv case:

- Buildroot overrides disappear on bodhi ticket submission:
When an override is scheduled for a period of time, it is discarded
when a bodhi updates is submitted. In my case, some dependencies
wasn't rebuilt yet, but was also FTBFS for other reason. Pushing an
update didn't invalidate the need to rebuilt this update with the new
ABI. That could have been done later and would still allows to test
the actual override.

- Buildroot overrides should be included in the dependency checks report.
That cannot replace an announcement to know if such ABI change is
legitimate. But that would help maintainers to keep in mind that their
action is needed and for which branch. Specially if they have missed
the announcement for any reason.

- About Bodhi and ACLs,
Tickets integrity shouldn't be affected by the ACL mismatch of each
single package by the submitter.
As I'm submitting an override, then I should inherit "some rights" to
push an update with consistency from testing to stable.
Probably bodhi should be aware that buildroot overrides are a premise
to an update and suggest to add the packages rebuilt with the new
version, and allow an ACL on it for the ticket.

I can understand that there is too much rights involved with the bodhi
buildroot override for non provenpackager. So it become a
provenpackager duty to drive an ABI change. That wasn't the case with
trac tickets because that has involved also some ACLs override to
rebuild a list of packages. And that was conduced in a single process.

Nicolas (kwizart)


More information about the devel mailing list