In a comment, Paul Snively wrote:
This is a good acknowledgement, but you also overlooked the entire field of Process Group Communication systems as exemplified by http://www.cs.cornell.edu/Info/Projects/Ensemble and http://www.spread.org. Spread even already has Python bindings; see http://www.python.org/other/spread.
To be quite honest, it's been rather frustrating to watch the Chandler development process as it seems to be developing virtually everything other than the GUI framework from first principles rather than leveraging the work that's being done by other organizations, particularly with respect to RDF storage and query and reliable decentralized networking. I can't help but think that a good RQL implementation on top of e4Graph, Secure Spread for networking, and KeyNote for trust management would have freed the team up to have made even more progress than it has, by enabling them to focus on application-level issues rather than nitty-gritty infrastructure.
But please take that with a grain of salt; I've been doing software long enough to know how easy it is for folks like me to be back-seat drivers. :-)
I appreciate your frustration. With 20/20 hindsight, we would no doubt have made some decisions differently.
We tried hard in every single respect NOT to develop things from first principles. And, in fact, we are building on far more major open source components than wxWidgets for the GUI. Our repository is layered on top of Berkeley DB and dbxml. We are using the Twisted networking libraries, Lucene for full-text indexing, and OpenSSL for security.
We evaluated lots of other alternatives, as long-time followers of this project will be aware. In particular, we evaluated a variety of RDF stores and found them wanting in terms of performance, reliability, and overall suitability for our needs. Unfortunately, not enough had been done when we made our choice a couple of years ago to move RDF into the practical world of software engineering. One of the overriding requirements was that any infrastructure we adopted should be sufficiently robust and scalable (or sufficiently close that we felt the gap could be closed, especially if we gave it a nudge) as to stand alongside any high-quality, commercially successful proprietary alternative without the least embarrassment or excuse. It's a high bar.
That said, my PC-centric background, and the normal fallibilities of experience and judgment of all teams contributed to how we got where we are.
Even more than this, though, and an area where I'd really have to agree is that I didn't grasp the schedule impact of decision-making based on an unwillingness to compromise key requirements. If I had... well, that takes us into speculation about alternate universes.
Posted by mitch@osafoundation.org at August 07, 2004 09:55 AMGood luck guys - hope you deliver something that manages to excite folks in the open source community and end users, before the end of the year. Otherwise you can kiss this project good bye !
Exciting folks in the open source community might not be as easy - with good alternatives like Ximian Evolution. The monkeys @Ximian came out with the initial release of Evolution pretty fast and kept up a breathless series of updates; today it has reached a point where it is ready to shed its outlook look and feel, and interoperates with Exchange. The key thing is development momentum - not getting some fluff pieces planted through friendly journalists.
Posted by: YABSD - Yet Another Back Seat Driver at August 7, 2004 11:17 PM
YABSD - I never ever had a chance to be more than a BSD.
Posted by: DaniGro at August 9, 2004 07:27 AM
Re: YABSD
It's easy to fault the Chandler project for not delivering something interesting/useful in a time frame that we all might have liked. That doesn't mean it won't eventually deliver big, and have a big impact.
If you got back enough years, you'll remember how long it took the Mozilla project to be anything other than a complete fiasco, schedule-wise and feature-wise. It took them far, far longer than hoped to deliver something useful to anyone other than the Mozilla developers themselves.
But look where they are today. Mozilla-based browser share is measurably growing, and by the millions. Their momentum -- among both Open Source advocates *and* the mainstream press -- is big enough that Microsoft has now reorganized the IE team, and reportedly changed their plans to wait for Longhorn to ship a new version of IE.
A long time saying in tech bears repeating: we overestimate change in the short term, and underestimate it in the long term.
Posted by: Alderete at August 11, 2004 11:11 PM