[lilycore-dev] Long time no see

[Will C. <wi...@co...>]

It's been a while since this list has been used for anything other than 
reporting bug fixes. Enjoy!

There's been a lot of talk on the RPI core in the past week about 
server development, lack thereof, and how hard it is to get anything 
done with "the server". I completely agree. There are technical 
barriers to entry. There are social barriers to contribution. We're 
(note: not /I'm/) going to mitigate as many of the issues as possible, 
leaving folks with the ability to actually contribute in a useful 
fashion as their time and skills permit. I must stress, however, that I 
plan on writing very little code to fulfill this prophecy. So far, in 
all the discussion I've seen one person say they'd contribute if we 
provided some infrastructure tools. So, here's hoping they deliver. =-)

(And please, gentle reader, before you say, "but there's more to 
development than coding", I know. I know. See followup emails for more 
details, and for the meantime, curb your pedantry.)

My laissez faire attitude has been (easily) misinterpreted as "no, we 
ain't doing that", which, ironically, is exactly the thing I was trying 
to avoid. In my effort to get out of the way, everyone seems to think I 
took the road with me. I'll attempt to fix that by putting up some very 
large green signs, and we'll see if anyone follows them.

BTW, anyone who is NOT on the RPI core [1], and is interested in doing 
development work, I apologize: there's been an even larger barrier to 
entry for you since we haven't been using the other online methods of 
communication available to us, just RPI core.  If you're not on RPI 
core and wish to contribute, by all means, join the mailing list, drop 
us a line, or request an account on the RPI core [2], and join us in 
the -lily-dev discussion.

** WHO WE ARE

Coke@RPI:

Christian@RPI is the primary architect. My job description used to be 
"thorn in Christian's side". Sadly, that's no longer an option. Since 
he retired, I'm more of a project manager, though I've not had many 
resources to manage.

Garance@RPI :

Garance is pumpking [3] for the 2.8.0 release. Also the person who owns 
the server RPI's core runs on.

Wakko@RPI:

In charge of the documentation.

Others:

Folks that are currently attributed to a problem report, at least 
according to Jitterbug [4]. Nautilus@RPI, Transparent@RPI, Ach@RPI, 
Christian@RPI. There are all the client developers, and the peanut 
gallery in -lily-dev@RPI. I apologize profusely if you're in the middle 
of some server work and I missed you. Please drop the list a line to 
let me know what you're up to.

** THEN

In the beginning, the production core at RPI was patched on the fly, 
and it was good.

In the middle, a devCore was created. Most development took place here, 
though some still happened at RPI. Cores were painstaking carved from 
the handcrafted devCore.

Near the end, our primary architect did a large amount of development 
taking a stock 2.6.4 core, fixing lots of bugs, adding lots of feaures, 
and providing the release to the community, to be integrated by the 
pumpking. (who, at the time, was Coke@RPI)

** NOW

This didn't happen (it was originally hoped that this would occur by 
the end of 2002). I got sidetracked, for both technical and social 
reasons. Garance@RPI thankfully pulled up the slack and is working on 
integrating RPI, devCore, and Christian@RPI's final dev copy. 
(Christian's did a substantial amount of development against a stock 
2.6.4 release)

Garance is, SFAIK, mostly done with this integration work. The 
outstanding issues include a few high priority jitterbug tickets, and 
help. Garance's work here will be used as the 2.8.0 cut.

Kudos to Garance for doing this much needed integration work.

** LATER

Although lily is a mature project, there is still work to do. However, 
once a project becomes "mature", there is a palpable amount of inertia. 
Things work (mostly).

I think one of the other biggest recent problems is a rather circular 
problem of "lack of volunteers" and "lack of direction". Given that 
lily is staffed by volunteers (who typically have other projects on 
their plate), there's a definite concern about investing cycles on 
potentially useless tasks. So, they may not bother writing up a list of 
things to be working on, or design notes, because there are no 
volunteers to implement them anyway, but instead hack on a particular 
bit of low hanging fruit, or fix a particularly shiny bug.

On the other hand, potential volunteers may be looking for work to do, 
but given a lack of direction, generally wander off, or don't speak up. 
( Not to mention lily's history of burning out contributers, which I'm 
sure it shares with other projects. ) Client developers, coming in from 
the side, occasionally have very good requests for updates which can 
easily fall by the wayside in large part for the reasons described here.

In the interests of furthering my secret plan, and breaking the 
apparent deadlock here, I'll present a roadmap of sorts over the next 
few emails. I stress: if you're really interested in lily server 
development, you're going to have to do some of the work. Hopefully 
we'll make that easy for you to do.

My next email be a partial listing of outstanding projects. I will 
followup with notes on a few of the most important items, all of which 
will be geared towards making work on the remaining issues 
distributable to those interested in working on them.

~Coke@RPI

[1] http://rpi.lily.org:7780
[2] http://rpi.lily.org:7780/new/
[3] http://www.nightflight.com/foldoc-bin/foldoc.cgi?pumpkin
[4] http://www.centauri.org/cgi-bin/lily
--
Will "Coke" Coleda                                    will at coleda 
dot com