EVR upgrade path check

Nikolay Vladimirov nikolay at vladimiroff.com
Sat Jan 24 21:59:28 UTC 2009


2009/1/24 Bruno Wolff III <bruno at wolff.to>:
> On Sat, Jan 24, 2009 at 15:42:20 -0500,
>  Christopher Beland <beland at alum.mit.edu> wrote:
>> So I just closed bug 476375, which concerned a problem with Fedora 11
>> Rawhide having a lower version number than Fedora 10.  But if someone
>> releases an update to Fedora 10 with a higher Epoch-Version-Release
>> number, this could crop up again, and I wouldn't notice it because I
>> only run the upgrade once.  I notice here:
>>
>> http://fedoraproject.org/wiki/QA/ReleaseCriteria
>>
>> that these EVR bugs are considered release blockers.  I would feel a bit
>> more comfortable about closing this bug if I knew that any future
>> manifestation would be caught by a systematic check.  I've heard mention
>> that there used to be a script that checked for this sort of thing, but
>> after all the infrastructure changes, it's offline.  I heard there was
>> some sort of controversy about what it should do; is there any movement
>> on getting that ironed out, or is it still stalled?
>
> I had filed an RFE (478697) about this a few weeks ago. James Antill posted
> a sample script as an attachment to the bug.
>
> --
> fedora-test-list mailing list
> fedora-test-list at redhat.com
> To unsubscribe:
> https://www.redhat.com/mailman/listinfo/fedora-test-list
>

Just to point that this is very important.
I always do a fresh install on new releases and after a while I enable
the rawhide repos and do a 'yum update' and this is very annoying
almost every time because of EVR bugs. There are numerous other use
cases where EVR bugs really annoy people a lot of them were discussed
in this list so no point to bring them out.
I don't blame the packagers since some maintain 50+ packages and it's
quite hard to sync all of them for rawhide, the updates-testing and
updates for the previous two releases. So some automatic check will be
very nice.

-- 
NV




More information about the fedora-test-list mailing list