Forum OpenACS Development: Status and Questions

Posted by Ben Adida on
Roberto asked some very pertinent questions on the "obsolete"
OpenACS mailing list, and I want to take some time to answer
them here on the bboards, as well as give you all an update on
the OpenACS 4.x port.

Roberto's main comment is that, while he understands that core
decisions remain a closely watched process, he doesn't
understand why the process seems so closed. I think it is only
fair for Don and I to face this criticism given that we've made the
very same comment to ArsDigita.

First, I would like to point out existing avenues for observing,
commenting on, and contributing to how things are being
developed with OpenACS 4.x. The DB porting recommendations
are up and you're welcome to add comments to them. Same
goes for the Query Dispatcher. In fact, I'm currently revising the
specs to take into account some very insightful comments made
by the community. I've begun to hack a bit on the Query
Dispatcher code itself, which you can find in the SDM Repository
in packages/acs-tcl/tcl/db-query-dispatcher.tcl. I *just* started
yesterday, so there are certainly many improvements to make. I'll
be continuing the hacking today and this week. You'll note my
extreme attempt at commenting everything I'm doing so that this
code is as clear as possible to the community at large.

It is true that Dan, Don and I have been holding a couple of side
discussions. Most of these, though, have been related to the
surprise release of ACS 4.2beta from ArsDigita (which we didn't
know was coming until we saw the bboard posting). We've also
found out that ACS 4.3 is in the works. Thus, the discussions
we've been having are mostly organizational: how do we deal
with new code from aD when we thought we were now on our
own path?

However, I can understand how this looks somewhat closed to
the community. Thus, I propose that we use the OpenACS 4.0
Design bboard to discuss *everything* that is going on. Don and
Dan, I'm putting you on the spot  to discuss these things publicly,
but I doubt that this will be an issue with you anyways, correct?

In terms of OpenACS progress, I think the important news is a
slight restructuring of the team. Whereas you saw me
coordinating the dev effort for 3.x, Don Baccus will be
coordinating/managing the 4.x effort. Don simply has more
experience than I do in coordinating large development efforts,
and we want to build 4.x the "right" way. Don also has a bit more
time than I do to perform the more constant and rigorous role of
coordination. Don and I will remain the "gatekeepers" of the code
and specifically of the data model. I will focus on overall design
issues, maintenance of the original ACS philosophy, continued
communication with ArsDigita, and specific coding efforts (query
dispatcher for now).

Dan Wickstrom, who ported large chunks in 3.x, is attacking
large chunks of the core data model, and Don will be pulling in
one or two experienced 3.x developers to help out with that job,
too. Dan Wickstrom eats SQL outer joins for breakfast, so we're
in solid hands there.

So, let's be even more open here. We are trying to accomplish a
tough job: keeping this a fully open-source project in a
community of very diverse opinions (porting to Python, porting to
Java...), while focusing on porting the core ASAP so package
hackers can feast on their porting goals,  too. Bear with us as we
try to improve our organizational process.

And, of course, if you have suggestions, please respond to this
thread and tell us what you think!

Posted by Don Baccus on
I responded to Roberto's e-mail in private, as well.  Yes, we need to
work harder to communicate.

One problem we faced this week, though, was that we were asked to keep
the information we were getting in regard to 4.2, a later 4.3 (which
is really a minor update), and the eventual fate of ACS Tcl at aD (NOT
HERE) uber-confidential.

I won't get into a public discussion as to whether or not keeping
information on the future of ACS Tcl at aD confidential is smart on
aD's part or not.  It's not our (OpenACS) call, it's theirs.

Many of you probably noticed a couple of web/db threads where that
future was discussed somewhat openly by aD'ers as well as other folks
in the community.  Jon Griffin started that thread, and by coincidence
it was a couple of days after the "confidential" threads and
discussions between aD and a couple of OpenACS folks and aD partners
began (I was included under both hats due to the fact that I do some
consulting for furfly).

So it was a personally confusing and frustrating week.

That's not a good excuse for our not communicating better, we could've
at least said "we're talking to aD about 4.2 beta and future versions
but have been asked not to discuss things publicly".  Probably
wouldn't've made most folks here happy but, again, keeping things
quiet is/was aD's call.

I'll be out of town from Wednesday afternoon (21st) to Tuesday
afternoon (27th) for a brief vacation.  Before leaving I'll e-mail
those who've offered to help port modules to get some discussion going
regarding who will do what, and what kind of help folks will want.

I also expect to get my changes to the bootstrap loader (to load the
right datamodels) committed to the CVS tree before leaving.  Dan
Wickstrom's committed his first cut at porting the datamodel to the tree.

And ... anonymous checkout via the pserver now works, as of last night
(Janine Sisk confirmed this after Ben fixed it).  If you want to take
a look at the work Dan's done this far, grab a copy.  There's a new
"postgresql" directory under the package "sql" directory for the
pieces he's got loading (the oracle datamodel will be moving shortly
to "sql/oracle", too).

Take a look.  Feedback's always welcome...