Re: review process not working

[Graham Percival]

On Tue, Jul 26, 2011 at 07:36:43PM +0200, David Kastrup wrote:
> 
> So it needs to tell its story in comments.  It doesn't.  There is a lot
> of code in Lilypond that nonchalantly expects people to get along
> without commenting what it does.  This is often a nuisance, but if the
> code is written by a master, the pain of figuring out what it does is
> usually tolerable.

Very true; I think that almost all developers (if not all!) would
like to see better comments in the code!  Could you get into the
habit of pointing out questionable / unclear parts of code in
future patches, so that the author will explain them in comments?

I imagine that the first few weeks would require an awful lot of
reviews to get a good level of comments, but over time authors
would get used to the amount+type of comments that we want to see.

> Masters don't grow on trees.

Indeed.  We need to nurture non-Masters -- speaking of both people
like Carl and Neil (who, although knowing an incredible amount
about lilypond, are still not perfect), and people like Janek and
Bertrand (who are relatively new contributors).

How can we nurture those very different types of people?
- more reviews
- reviewing their reviews

For new contributors, a mentorship system would help, but that
requires mentors (which are in short supply).

[the patch author]
> invested more care and effort than he sees us doing.
> 
> Can we do better?

Yes.  Do more reviews, and be more careful in reviews.  If I could
wave a magic wand, I think the first thing I'd ask is for all
developers (i.e. people with git push ability) to spend 1 hour a
day reviewing patches.  That's obviously not going to happen, but
let's stumble forward as best we can, with the resources at our
disposal.

Cheers,
- Graham