Forum OpenACS Development: OpenACS 4.1 port update/status

Collapse
Posted by Don Baccus on
OK, some things have firmed up ...
  • We will NOT be using PostgreSQL's inheritance features in the port. After giving it a lot of thought, reading commentary to the thread here at openacs.org, and hashing things out with Ben, we've somewhat reluctantly decided that the current PG implementation of ORDBMS features carries too many negatives to make use of it a good idea.

  • Dan Wickstrom's started work on the core datamodel and is making progress.

  • Ben's going to try very hard to get his query dispatcher draft spec ready for comment by the end of this week, and you're all invited to toss snowballs at him afterwards (though the snow might melt by then!)

  • I'm making good progress getting the bootstrapper and installer to be "RDBMS-aware" rather than "glued to Oracle".

Roberto - you should contact Dan and get him to supply you a copy of the first PL/SQL package (or snippet thereof) he gets ported over for your PL/SQL Package port example in your PL/SQL->PL/pgSQL document.

Those of you who are interested in porting modules should e-mail me privately and/or post here. I'm going to start collecting a list of folks interested in pitching in. We need folks willing to help with testing (PG and Oracle) and documentation as well as porting. If you're interested in porting a particular module, please let me know.

Once the query dispatcher is spec'd and a prototype running we should be able to make fast progress on the core. Work on the core at the moment will be done by Ben, Dan and I, partly because there's just not enough there to divvy up into smaller chunks once we get rolling.

An immediate task someone could pick up would be to rewrite the porting guide for PG 7.1. Ben, Dan and I can help with specifics if anyone is interested in picking it up. For instance, the whole section on converting outer joins to (ugh) UNIONs needs replacing with a guide to converting Oracle to SQL92 (and PG) style outer joins.

When I have a better idea of how many people are interested in actively helping out, we can start talking about just how we're going to split this job up.

Collapse
Posted by Roberto Mello on
I just committed the PL/SQL-to-PL/pgSQL-HOWTO to the sourceforge CVS tree. The PG team will include this in the main docs and they gave me until March 16th to submit a revamped PL/pgSQL documentation. I will be working on it next week (spring break :)).

I can work on the content repository as soon you guys give us the "go" signal.

How is CVS access going te handled?

Collapse
Posted by Ben Adida on
CVS access information will be posted next week. We are going to run our own CVS, using the SDM CVS access feature and patch management to get a real community development system going. Get psyched!
Collapse
4: Photo Album 4.x (response to 1)
Posted by Walter McGinnis on

I'm interested in porting the Photo Album.

Collapse
5: Porting question (response to 1)
Posted by Erik Rogneby on
Is it the general consensus that porting ACS 4.0 packages to the OpenACS is better time spent that packagizing OpenACS 3.2.4 modules?

My first impression is that this is an almost rhetorical question.  But then I am wondering if there is some 3.2.4 functionality that will fall through the cracks... (also Civilution has done a pretty robust port of the photodb code that we'll be putting out soon)

I think I personally am up for porting the Spam System and the e-mail handler (and ACS Content by dependancy), and Civilution (me plus 4 other OpenACS developers) would probably ante up and tackle Templating.

After looking at http://www.arsdigita.com/acs-repository/ I find myself wishing for a dependancy tree.  For example, Photo Album required Templating and Content Repository.

Where are folks needed the most?

Collapse
Posted by Don Baccus on
Boy, this is a good question ... I look at ACS 4.2 as being much less
mature than 3.4.  I also think it is a more solid base for future
development.  There's a lot that needs doing to make ACS 4.2
reasonable for deployment, i.e. a decent admin UI for starters.

A dependency graph would be nice, and a tool to generate dependency
information even nicer - the APM should provide this, I guess you're
saying it doesn't?

As far as where we'll need the most work, folks are starting to e-mail
and post with offers of help so in a few more days I should be getting
a better handle on what we've got covered and what's uncovered in
terms of getting basic stuff up.

Collapse
Posted by Michael Feldstein on
Speaking of a decent admin UI, I'd like to revive the idea of creating a usability testing module and suggest that creating one be considered as part of the OpenACS porting process. I know that the development team already has its hands full just porting and debugging the stuff that's already in ACS Classic, but one of the important contributions that the OpenACS project makes is better interface design. I think we could evolve the usability of the new system much faster if we had a mechanism for collecting usability data tied into the SDM and/or ticket tracker.

I can't do the programming myself, but I'd be happy to work on the design with somebody. I have an old spec I wrote kicking around my hard drive somewhere; it wouldn't be too hard to find and update it.

Collapse
Posted by Talli Somekh on
I'm interested in porting the Problem Sets to OpenACS.  It doesn't seem like it's so efficient to have new people learn the ACS on Oracle then try and port their brains to Postgres.  I know that this probably isn't such a tough thing to do compared to a query dispatcher, but I imagine there would be some interest in the community for PG problem sets.

Am I late on this?

tal

Collapse
Posted by Don Baccus on
Not at all, and even more interesting would be a comparison of
solutions as a way of comparing PG and Oracle...
Collapse
Posted by Ken Kennedy on
1)I'd be interested in porting the Bookmarks module.

2)And was there any clarification on anonymous CVS access yet? I keep my own CVS repsository (so I can track personal package changes [I have my own OpenACS3 "weblog" module that I want to port], other projects, etc.). It's so easy to do an anonymous checkout into /tmp, import that code into my repository, and then update my development area. Are we going to keep that sort of functionality?

Collapse
Posted by Don Baccus on
I'll put you down for bookmarks.

One of the openforce folks was working on getting pserver stuff up the
past couple of days but it wasn't quite right.  Janine Sisk dug in and
figured out what needs doing, I believe.  Should be up soon.  At that
point you'll be able to grab the tip of the CVS tree anonymously.

Collapse
Posted by Ken Kennedy on
Excellent!! I'll keep an eye out here, and go back and start re-reviewing the ACS4 doc stuff.
Collapse
Posted by Robin Felix on
In response to Erik's porting question:

There are people who eagerly climb the mountain "because it's there." There are others who just climb the mountain because they happen to live on the mountain.

The first group built OpenACS and are ready for the next challenge: 4.1. God bless you all for the trail you've blazed for the rest of us and success in your future endeavors.

The second group, including me, generally support a large, existing user base and installation. They will continue to work in 3.x.x and need a darn good reason before they move, because even a small move such as that from 3.2.4 to 3.2.5 becomes tremendously more complicated when there is a service-level to maintain. I loaded up ACS 4.0 (TCL, Oracle) on a test machine to check it out; I saw reduced functionality in several areas, notably in the delegation of group administration.

Bottom Line: There will continue to be a market for the 3.x variants until 4.x surpasses its functionality, regardless of its elegance in design. Where time is better spent is a matter of perspective.

Sub-Bottom Line: I will continue to submit trouble reports and suggested fixes for the benefit of my neighbors on the mountain, and I look forward to any increased 3.x functionality (photodb-lite, anyone?). 😉