Packages updates on REVU

[Lionel Le Folgoc]

On Sun, 22 Apr 2007 13:35:15 +0200, Gauvain Pocentek <gauvainpocentek at gmail.com> wrote:
> Hello MOTUs and MOTU hopefuls,
> 
> As you certainly noticed, the list of packages waiting on REVU is
> really, really long. I've reviewed a few and the update to gutsy is
> needed for all of them (only a one line change in the changelog, but
> needed). It'd be nice if  people having  packages on REVU could update
> the distro target.
> 
> Also, a lot of packages have been commented by reviewers but not updated
> by the uploaders. Could we define some kind of policy about what we
> should do with this packages? My guess is that we should mail the
> uploader directly (not only a message on REVU or on this list) to ask
> him what he intends to do with the package. This would allow us to clean
> the list, and other packagers to grab the package and finish it if the
> initial packager don't want / can't work on the package anymore.
> 
> Thoughts?
> 
> Cheers,
> Gauvain
> 
> --
> Ubuntu-motu mailing list
> Ubuntu-motu at lists.ubuntu.com
> Modify settings or unsubscribe at:
> https://lists.ubuntu.com/mailman/listinfo/ubuntu-motu

Hi,

I think this is a good idea.

But IMO, it raises some important questions, and since we're at the
beginning of a new development cycle, it is the good time to try to
answer them, to keep on improving us and Universe.

I read some comments on bug reports[1], which made me think that
something isn't going well, and we should try to find why and how fix
it. And the thing to blame isn't 'miscommunication'.

Yes, everyone will agree, the packages list waiting on REVU is huge, and
I believe there is an issue in both MOTUs and people proposing packages.


* First, why are there so few reviewers?

That's not definitely because a lack of MOTU (65 members of ~motu)
I really wonder why so few MOTUs are reviewing on REVU. Here are some
possible answers:

 - core-dev: some MOTUs are also core-dev, so they are very busy,
dealing with main packages.

 - "debian-way-of-thinking" (that's not pejorative): some MOTUs take
care of their own packages *only* (that's not the way I see this task,
and I am pretty sure you agree).

 - too busy to review: making statistics, rebuilds of the whole archive
and all this kind of tasks is useful, but not essential. It won't
fix nor review issues/packages.

 - knowledge issue: reviewing is hard, and reasonably, no one can know
all packaging policies. I think that many MOTUs don't 'dare' to review
package, because they think it'll call into question their packaging
knowledge.


* Then, why are there so many packages waiting on REVU?

(I've just spoken of the lack of reviewers above, so I won't speak about
it again.)

IMO, many packagers upload their packages to REVU, and think that's
done, and their packages are going to get into Universe. I don't know if
it is a miscommunication or a lack of motivation, but actually, very
few people come on #ubuntu-motu to "prevent MOTUs from sleeping until
they have reviewed their packages". This is worse during REVU days: many
packages are reviewed, but few are updated by packagers within the "time
frame" (i.e. before the end of these REVU days).


* So, how can we improve this?

I think the REVU/wiki page has to be updated, to explicitely say:
"When-you-upload-a-package-go-on-#ubuntu-motu-and-cry-and-yell-until-you-get-a-MOTU-to-review-it".

Besides, we need a better tracking for proposed packages. The
motu-reviewers mailing list has a 'too high traffic' to manage this
efficiently. When I review a package, I need to keep the name/address
and check incoming mail to see if it hasn't been updated.
I think we can use LP for this. For instance, if a bug report is filed
for each new package uploaded on REVU, it would be easier to see if a
MOTU isn't already reviewing it. As we did with the 'needs-packaging'
tag, we could set up a tag which indicates a package needs to be reviewed.

Finally, I think that it'll only need a little effort if everyone is
involving ; no specification - even well written -, no statistic - even
very precise - is going to be the miracle-fix for the 5 problematic
points explained above.

Thoughts on this?

Cheers,
Lionel

[1] https://bugs.launchpad.net/ubuntu/+bug/104654/comments/2
    https://bugs.launchpad.net/ubuntu/+source/vdrift-data/+bug/106854/comments/2