Contributors mailing list archives

Browse archives


Re: Self-merging rule

Camptocamp France SAS, Alexandre Fayolle
- 31/08/2015 16:47:20
On 31/08/2015 13:53, Leonardo "LeartS" Donelli wrote:
> The self-merging is a symptom of the disease, i.e. not enough reviews
> and reviewers.
> In theory, not allowing self-merging is a good thing, for all the
> reasons already explained by others.
> In practice, at the moment there are:
> * very few occasional reviewers (probably less than 30-20?)
> * even fewer somewhat "consistent" single-repository reviewers
> * 1-3, consistent cross-repository reviewer, Pedro.
> * 925 open pull requests
> * ~100 open PRs for new modules (rough estimate based on title filtering)
> * 462 open PRs created before June 2015
> Of the 925 open PRs:
> * pedrobaeza has commented on 429
> * gurneyalex has commented on 139
> * yvaucher has commented on 124
> * dreispt has commented on 64
> * eLBati has commented on 61
> * guewen has commented on 59
> * nhomar has commented on 30
> (excluding PRs opened by the user himself)
> Then there are a dozen or so people at 10-20, after that..
> All of this top reviewers are also top contributors, so it's easy to
> see why self merging is "required" at the moment.
> I'm not saying we should all get to Pedro's levels, but there is
> indeed a problem with the contributors vs committers vs reviewers
> balance.

Hi LeartS

I agree with your diagnostic, although the situation is much worse than
what you describe, as you only consider open PRs. I expect that taking
also closed PRs into account will show an even greater spike for Pedro.

My opinion is that Pedro should relax a bit, because at the moment lots
of people are relying on him doing all the work, and this is giving them
(both PR submitters and members of the committers team). :-) (I
seriously mean it).

I try to make 3-5 reviews for each PR I submit, and this is already very
time consuming. If every PR submitter took the time to do the same, the
overall situation would be very different and the core committers could
certainly focus on doing final reviews and merge, which would be much
more efficient. I'm tempted to start asking people proposing new modules
to show me the list of OCA PRs they've recently reviewed and bluntly
refusing the PR if they are unable to show me the reviews (bug fixes and
ports are a slightly different topic in my eye as it qualifies as
community service, but still).

I'll remind everyone of the existence of the #oca IRC channel (freenode)
which could use more lurkers. It can also be used to gently ask for a
review on the PR of your heart which has gone below the radar, and the
active members of the chan usually comply in a very timely fashion. This
can be a good way of avoiding the 'self merge' temptation (at least if
you leave in a western european TZ. If you don't, then we need to expand
our commiter team to people in remote TZ, and in the meantime, you can
gently ask here if your PR has not gotten reviews / merges after several

(an important key word in the previous paragraph is "gently" :) )

Alexandre Fayolle
Chef de Projet
Tel : +33 4 58 48 20 30

Camptocamp France SAS
Savoie Technolac, BP 352
73377 Le Bourget du Lac Cedex