Re: Press Release -- Numbers

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


Forgive the reposting: my mail program had an error that may have
caused some of you not to receive these 2 messages I sent earlier.
Thanks to Robert and Justin for bringing it to my attention.

Message #1:

> Discussing the press release on IRC, we realized that we really want
> the number of code, /contrib, and GBORG contributors. Can anyone think
> of a way we could get that?

That's a tough one. One way would be to scour the CVS logs: the
core maintainers usually attribute who a certain change was from.
How you could automate that is beyond me: I have some ideas, however.

Another way is to look at the email lists. Since most patches come
through on the patch list, I did a quick count of distinct "from"
addresses from that list. My archives only go back about a year:
in that time, there were 1814 messages from 94 different emails.
About 24 of those were one-shot wonders, but the usual suspects
topped the list:

    73 | Neil Conway <neilc@samurai.com>
    83 | Peter Eisentraut <peter_e@gmx.net>
   129 | Joe Conway <mail@joeconway.com>
   307 | Tom Lane <tgl@sss.pgh.pa.us>
   599 | Bruce Momjian <pgman@candle.pha.pa.us>

So I think about 90 is probably a good ballpark to start from,
as far as the number of people contributing to the code. Assuming
that almost all of the posters to patches are actually contributing
something. (By way of comparison, the general list saw @28,000 posts
from 3200 people!) This is for one year, so a total I would roughly
guess to be about 2-3 times that. I'll go see what I can do with the
cvs logs...


Message #2, some time later:

> Discussing the press release on IRC, we realized that we really want the
> number of code, /contrib, and GBORG contributors.   Can anyone think of a
> way we could get that?

I parsed through all the cvs logs (256,000 lines) and came up with a total
of 573 distinct names that were mentioned somewhere in the comments. This
does not count the actual cvs committers, but they all mention each other
at some point. Very unusual names may have been overlooked (e.g. no caps
at all in the name), but I'd say this is a very accurate count. I have all
the info in a database now if anyone is curious about any other stats.


Greg Sabino Mullane  greg@turnstep.com
PGP Key: 0x14964AC8 200211160722

-----BEGIN PGP SIGNATURE-----
Comment: http://www.turnstep.com/pgp.html

iD8DBQE91jkPvJuQZxSWSsgRAjSwAKCFBBbLOPmQPj8TjvuoW8ytXYhHwQCg4QNm
dEY4BkvIIFAqc0lNJOIXtXo=
=Qmpn
-----END PGP SIGNATURE-----