What a bad example [LWN.net]

What a bad example

Posted Jan 9, 2008 21:41 UTC (Wed) by BrucePerens (guest, #2510)
Parent article: Open Source Applications Foundation restructures

OSAF was a classic example of how not to do an Open Source project, by applying big-company software engineering practices and spending lots of money, over lots of years. I'm glad they have something to show, and hope the project can attain a real community with a different paradigm.

Somehow a number of the same people managed to get the Mozilla foundation going, which is a much bigger achievement (although it also has some big-company problems now).

Bruce

to post comments

Mozilla

Posted Jan 10, 2008 2:10 UTC (Thu) by roc (subscriber, #30627) [Link] (1 responses)

It's hard to serve 100M users and get a lot of things done without *some* big-company
problems.

Mozilla may have (had) some board overlap with OSAF, but there's never been any real overlap
in the driving forces behind the organizations.

Mozilla

Posted Jan 10, 2008 16:32 UTC (Thu) by BrucePerens (guest, #2510) [Link]

Well, Mitchell Baker and Mitch Kapor were the overlap, that's pretty significant.

What a bad example

Posted Jan 14, 2008 3:26 UTC (Mon) by emk (subscriber, #1128) [Link] (1 responses)

Well, at the beginning of Chandler, I spent an afternoon reading their design documents. Even
in the commercial world, I've seen very few organizations dedicated to such generalized and
abstract software.

This is a pretty risk path to take--it's nice to have a clean architecture, but if you can
make enough simplifying assumptions, you'll get a product out the door much sooner.

What a bad example

Posted Jan 14, 2008 21:41 UTC (Mon) by hazmat (subscriber, #668) [Link]

i've been watching the project since its inception, have contributed several patches. i think
its more than just the architecture abstractions, although that was definitely a strong sign
of some of the issues. it was also poor technology choices (wx, chandlerdb) in addition to
poor management that lead to this road, imo. and inexperience with python, mucky internal api
(verbose xml initially).. it had the feel of abstracte java apis in python.. much of which was
cleaned up at with philiby eby working (peak, setuptools) as a contractor with them, but the
writing was on the wall.

in general watching chandler and osaf, was like watching a slow moving train wreck.

there server side story is much better (servlet cal-dav repo w/ sharing and  web ui), and was
built fairly rapidly by a small team.