Re: GOP-PROP 2: mentors and Frogs - probably decision

[Graham Percival]

On Wed, Jun 22, 2011 at 12:11:36PM -0600, Carl Sorensen wrote:
> On 6/22/11 10:09 AM, "Jan Warchoł" <address@hidden>
> wrote:
> 
> > I don't want to suggest laying implied obligations on anyone.
> > Maybe there's a misunderstanding, so let's make things clear: do you
> > say that asking Mike (who wrote automatic beam collision avoidance)
> > "Mike, I'm trying to improve automatic beam collision avoidance, but i
> > can't figure out what foobar does. Can you give me some clues?" is
> > laying implied obligations on him?
> 
> I think that the implied obligation is saying "Mike will help people who are
> playing with beam collision avoidance."

Perhaps the word "expectation" would better express my worry.

Suppose Janek says "hey, I want to work on frets" to -devel, then
Graham repliesr to say "great!  Go talk to Carl, he's the expert
on them".  Then Janek emails Carl personally (maybe cc'd to
-devel, maybe not).

1. will Carl feel that he should reply because Janek cc'd him?
2. will Carl feel that he should reply because Graham said,
publicly, that Janek should talk to him?
3. if Carl is busy and ignores the cc'd email, will Janek feel
disappointed?

I want to avoid all of those.  At the moment, I see little
evidence that emails to -devel are routinely ignore by people who
would want to answer them but just didn't notice that email.  In
the past 3 years, I can think of perhaps a dozen cases where this
happened -- but with the PATCHES: 48-hours   emails and the
GOP-PROP  emails, I think the danger of this miscommunication is
much lower.


I know that you think that new contributors should get more help,
but I think we've got as much as we're going to get.  Every
developer who offered to mentor has a contributor.  Many of them
are "experienced contributors" who will probably become
fully-fledged developers in a month, and that will hopefully allow
us to accept give a few more contributors dedicated mentors.

Think about it this way: at the moment, we don't even have a
smooth process for developers.  There's 33 outstanding patches,
many of them by developers.  15 of them have been abandoned --
about half of them were abandoned because they were silly ideas,
but the other half are totally doable if only we have a bit more
discussion about how to finish the job.  There's another 15
patches that currently need work... maybe indentation, maybe a bit
more architecture rethinking, maybe stuff that's actually been
abandoned but nobody's bothered to mark it as such.

> >>> Problem 2: an Enthusiastic User writes to -devel: "it would be great
> >>> if Lily supported foobar!". Graham replies: "sure, add it yourself."
> >>> and nothing else happens. Graham is right, but user gets discouraged.
> >> 
> >> Is that a problem?  :)
> > 
> > We don't know, but probably it is. You said that Carl started as a
> > documentation writer, so maybe that Enthusiastic User would become
> > next project leader, after 3-4 years?

There's a 20-30% success rate for turning new contributors into
developers.

> > You wrote yourself that recruiting new developers is crucial for the 
> > project.

Long-term, yes.  Short-term, no.  I think the short-term priority
should be making sure that development process is working for our
main developers -- because at the moment, we're not even getting
that much done.

That said, I want to treat people honestly.  If that means
discouraging some new contributors, then let's do that.  Hopefully
in 3-4 months we'll have the development process sorted out for
developers, and then we can consider actively recruiting new
contributors.  But it really will be months before we're in any
position to support such a recruitment effort.

> > A question: are there (m)any people in the development team who
> > started as developers right away, without being "just users" for some
> > time? (besides Jan and Han-Wen, of course)
> 
> Actually, I started as a developer right away, because I wanted guitar fret
> diagrams.  It was only after that development that I ended up helping with
> the documentation work under GDP.

True.  In fact, a quick glance at the paconet lilypond stats
suggests that out of the top 30 developers, only about 20% of them
(including me) started off as "just users".

Cheers,
- Graham