REVU - Cleanup of the "Needs Review" section

Jordan Mantha laserjock at ubuntu.com
Thu Nov 6 15:35:58 GMT 2008


On Thu, Nov 6, 2008 at 4:00 AM, Michael Bienia <michael at vorlon.ping.de> wrote:
> On 2008-11-05 16:30:19 -0800, Jordan Mantha wrote:
>> On Wed, Nov 5, 2008 at 3:17 PM, Siegfried-Angel
>> <siggi.gevatter at gmail.com> wrote:
>> > Jordan: The intention of this is not to let the list look better, but
>> > to don't waste time reviewing the packages of those people who we have
>> > already lost and who may not come back again, but instead to focus on
>> > the packages of contributors who are still active.
>>
>> I thought the point of REVU was about getting packages into Ubuntu,
>> not filtering out people who got discouraged.
>
> Yes (to some point). But I would prefer if we let only those packages in
> where the contributor is also interested in the package after inclusion
> into the archive. We don't need more packages which nobody doesn't care
> about afterwards.
> According to http://qa.ubuntuwire.com/multidistrotools/universe.html we
> have 851 packages in universe that are not in Debian sid. Let it be 500
> packages after filtering out the language packs, translations, etc.
> Who is going to maintain them? Certainly not MOTU which is busy enough
> already.

Well, the amount of packages not in sid is sort of where I'm heading
here, thanks for bringing it up. It seems fairly clear to me that
somebody doing intrepid->jaunty and reuploading is not much more
likely to maintain the package after inclusion. REVU is a huge problem
in the sense that we put packages in Universe *assuming* they will be
maintained, but they mostly just aren't. My argument has been that the
intrepid->jaunty thing is rather trivial and arbitrary. It's like
rejecting a debdiff in the sponsorship queue because of it. If we are
going to filter packages, let's do it right, let's really figure out
who is going to maintain their packages. Perhaps we need some sort of
unofficial contractual thing, where people put their name as Uploader
or similar in the package and when there's a bug, etc. we can pester
them :-)

> We certainly shouldn't add more unmaintained packages to Ubuntu.

Totally agreed, but I don't think this is exactly a good method of doing that.

> Besides this it doesn't help reviewing packages where the contributor
> lost interest in the package. Who is going to fix the package if not the
> contributor?
> We shouldn't spend the rare resource of reviewers on "dead" packages.

Agreed, but to me it seems that sending people this messages is just
further perpetuating the common notion that Ubuntu developers seek
trivialities to reject rather than face hard work. This is seen in
bugs, etc. If we're after prioritization, then lets do it, but let's
do it based on something less trivial. Many contributors don't see why
it matters what the target release is as long as their package is
good. To be rejected, especially if it is their *only* feedback, based
on that is sending the wrong message, IMO anyway.

>> That seems reasonable, but I wouldn't base it on something so trivial
>> as intrepid vs. jaunty. I'd think something like packages who haven't
>> had a comment from the uploader is so much time. We certainly should
>> prioritize packages that have received *no* review.
>
> So you would prefer reviewing packages where nobody is interested in the
> review anymore?

No, certainly not. And I personally think we should be removing some
packages from Universe that come from REVU for the same reason. I
guess my point is maybe better explained by an analogy to bugs. I see
this proposal as basically the same as closing somebody's bug with a
form "This bug hasn't had any activity for a while, I'm closing" when
no developer has even commented. *If* we are to reject bugs or
packages, surely it should be on the merit of the bug/package, not on
lack of activity on *our* part.

> If I would have time for reviews again I would concentrate on reviewing
> packages where I know that the contributor is still interested in a
> review.

It somewhat depends on what you're trying to get out of REVU I think.
If you're after good packages then contributor interest really isn't
all that important. Some people might just drop off very good
packages, but have no interest in maintaining them. As we have a team
maintainership policy I don't see an obvious reason to reject that.
Even more, maybe the MOTU is motivated to finish off a
decent-but-not-quite-there package and upload without the contributor.
However, if you are looking at creating contributors->MOTU then yeah,
the contributor is actually more important than the package they are
contributing. I'm just trying to point out that there are a couple
ways to look at REVU, IMO.

-Jordan



More information about the Ubuntu-motu mailing list