Re: Boston Summit 2006 is on Echelon's radar

[Shaun McCance <shaunm gnome org>]

On Sun, 2006-10-15 at 14:10 -0400, Luis Villa wrote:
> I know a couple folks were talking about something like this a few
> months ago, and I believe some code was even written- Shaun? There
> were even some less creepy names- heartbeat, maybe? :)
> 
> [Speaking as someone who wants to observe GNOME from afar, I'm
> intensely interested in seeing this go forward, so good luck.]
> 
> Luis

I called it Pulse.  Then the polypaudio went and renamed their
project to PulseAudio, so maybe that makes a naming conflict.
Maybe it doesn't.  Here's the rundown and status:

Pulse started out from Danilo's damned-lies (which is in CVS
right now), but I went in a much different direction  with it.
Its primary and original purpose is to report on documentation
and translation.  But once you've got the infrastructure for
all that, you can do so much more.

Pulse is a big pile of Python that collects information about
Gnome and displays it on the web.  It consists of a back-end
tool that knows how to collect data and some CGI that knows
how to display that data.

On the back-end, Pulse knows how to read some basic data from
some XML configuration files (which would live in the Pulse
CVS module) and do stuff.  Those files specify which modules
and branches we care about and a little information abut those
module and branches, such as the who the maintainers are.

Pulse keeps a database synced with the information in the XML
files and digs around for more information.  In particular, it
does full CVS checkouts and finds stuff like documentation and
translation.  I would also like it to find programs for modules
with multiple programs (e.g. gnome-games).

A general goal is to remove as much information as possible from
the XML files.  For example, when I first started out, I listed
all the documentation for each branch of each module in the XML.
Later, I made Pulse able to find these on its own.  I would love
to get maintainer and mailing list information automatically.

Pulse uses SQLObject for its database stuff.  I made heavy use
of InheritableSQLObject to make a clean and simple ontology.
Then I realized that vertical inheritance involves an insane
number of round-trip queries to the database.

The database design was just too slow, so I decided it had to
be replaced.  And that's when it stalled.

Basically, I'd love somebody who's good at databases to sit
down with me and flesh it out.  Then I'd love new contributors
to hack at it.  It's really fun to work on, and it lends itself
well to multiple contributors.

I never put it into CVS because I don't want to leave another
orphaned pile of unfinished code in CVS.  Honestly, our CVS
repository is starting to look like SourceForge.  But if we
have people interested, I will totally commit it and set up
a virtual hackfest.

--
Shaun