On Mon, Nov 11, 2013 at 1:51 AM, Tim Starling <tstarl...@wikimedia.org>wrote:

> On 08/11/13 03:40, C. Scott Ananian wrote:
> > Basically, every major piece of WP should have a module owner.
>
[...]

> > Certain people 'own' larger collections of modules -- like there are
> > subsystem owners in the linux kernel dev world.  For example, ideally
> there
> > should be someone who owns "the WMF deployed mediawiki" who can weigh in
> on
> > changes which affect the configuration and collection of modules which
> > actually constitute wikipedia.  And then there are the big three
> > ("architects") who are really just the top-level module owners (the Linus
> > Torvalds, if you will).
>
> My concern with this kind of maintainer model is that RFC review would
> tend to be narrower -- a consensus of members of a single WMF team
> rather than a consensus of all relevant experts.
>

To clarify, I wasn't actually suggesting that RFCs would be reviewed by the
minimal set of module owners.  The opposite, actually: I think that
explicitly creating a hierarchy of ownership would allow the review process
to efficiently progress from narrow to broad focus, ensuring that neither
end gets short shrift.  The line down the tree from "big three" to "person
who last touched a particular file" is a way to capture everyone who is
relevant to an RFC, making sure that neither the forest nor the trees are
neglected.  The main thrust is to flesh out the middle levels of the
hierarchy, lieutenants who have a moderately broad focus and who are
trusted to offload some of the work from the top 3 architects.  The top
three would still weigh in, but hopefully they can concentrate on the
broadest scale issues and wouldn't have to do as much of the heavy lifting.

I'm not proposing that RFC review should be done solely by the narrow-focus
people who happened to last touch the affected files, obviously.

That said, this is a relatively minor point; it seems we've reached good
consensus regarding the bigger picture decoupling of architectural
responsibilities and job titles.  Quibbling over the number and scope of
'architects' can be deferred (esp since the big 3 don't seem to be loudly
complaining of overwork at present).
  --scott

-- 
(http://cscott.net)
_______________________________________________
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

Reply via email to