I'll do these backwards, I think :)
6. aD Java appears to be dead, unless ex-aDers involved with the project decided to rally 'round it and finish it up. I don't know this for sure, am just reporting on what I've been told, which isn't necessarily accurate.
5. I think it's up to ybos as much as anything. We've encouraged them to play with OpenACS 4, since our core runs legacy ACS 4 Classic packages unchanged for the most part, and with minor changes otherwise. Those of us who've worked hard on the OpenACS 4 core would want an opportunity to review those changes before accepting them, of course, but forking two ACS 4 Tcl versions for Oracle is senseless. For instance, dotLRN features a very, very nice, brand-new portals package much better than either of aD's attempts (3.x or 4 Classic).
4. If you install OpenACS 4 Oracle it will seem very familiar to you. There are still some missing pieces in the PostgreSQL version. We're within a week or so of issuing a beta and will work on better documenting such things by release.
3. Often folks ask around here or by e-mail first. Outright bug fixes can be submitted by patch without discussion, of course. Cosmetic changes are subject to an individual's taste so worth discussing. As you've said personalization stuff's less likely to be of global use.
Being able to configure an optional "anonymous coward"-style user is probably something we should add to OpenACS 4 in the future. With a better permissions admin UI it would be easy enough to allow this user to post to bboards, etc.
2. Probably what you say, yes, though others may have cleverer ideas.
1. I'm not quite sure what you mean as it seems you started with 3.2.5? The fact that you used PG 7.2 is fine and we would like any changes needed for it to run with that version (though we'd like to see them run with PG 7.1, too). The parameters stuff you've fixed would be nice to have, too. Your note doesn't make it clear why you needed to change fips_county_code from char(5) to varchar(5), was PG 7.2 giving an error of some sort?