Forum OpenACS Q&A: Full text search in PostgreSQL?

Collapse
Posted by Guan Yang on
There is a full text searching module in the contrib section of
PostgreSQL. Are there any plans to integrate this with ACS/pg?
Collapse
Posted by Don Baccus on
Yes and no.  The full text searching module you're referring to is extremely limited so not directly useful to us.  The basic notion - build an index table to search - is the way one would have to do it, so the code might be poachable and serve as the basis of some more flexible and useful.  The guy who did it needed to find proper names, so does no stemming of words, doesn't attempt to deal with phrases, doesn't do any weighted matching, etc.  He did a great job of writing something to meet his personal needs, but no more.  That's probably why it's in contrib rather than mainstreamed.

I'm personally interested in doing a full text searching enhancement in the future.  Perhaps in summer?  It will depend on many things, i.e. contract work I might take on, the work I need to do on my own personal site, and how much Postgres internals work I decide to tackle  for 7.1 (I'm currently signed up for a small piece but may offer more  time).

Collapse
Posted by Guan Yang on
Perhaps you could use code and/or ideas from ht:/Dig or SWISH++?

Also, is there some way I could help fund this work? I.e., if I give you $x, will it get completed faster?

Collapse
Posted by Michael Feldstein on
Guan's question raises an interesting possibility regarding OpenSource business models. Right now, what I mostly see happing is individual clients paying companies like aD to build to their own needs. Then aD rolls whatever might be useful to everyone back into the code base. Alternatively, hackers add whatever they personally need to the code base.

It would be interesting, though, to set up a consortium where groups of users who want particular features could pool their money to pay the primary maintainers of the code to implement those features in the common code base. Ben, does this sound like a model that would be compatible with OpenForce.net? And is this something that others are doing now?

Collapse
Posted by Ben Adida on
Yes, it is an interesting question :) There are other companies doing this. Collab.net's SourceXchange for example. Or CoSource. However, I'm certainly not opposed to having our own process for OpenForce specific work. Anyone up for writing the bidding system for it? ACS/pg-backed, of course.

I think this kind of biz model is perfectly fine and should be encouraged, but it's not the main model I see as useful to the Open-Source community and the software industry at large. OpenForce is focusing on all of the services that go along with large software like ACS/pg. That means installation/configuration/support/customization (customization to a certain extent, this is not the ArsDigita model). Today, big companies spend 80% of their software budget on non-licensing issues. The open-source model, with its inexistent licensing costs, works beautifully in this context.

This may be as good a time as any to begin my call for help. OpenForce is growing, and we're working on our first few clients. Our model will include a distributed service force that should be made up of good hackers like all of you ACS/pg'ers. So, if you're interested and have at least 10-15 hours/week of time to put into this, please get in touch with me via email.

The software OpenForce will support will almost always be based on ACS/pg (and sometimes ACS/Oracle, although we will just be providing the interface to Oracle, not the Oracle help), with more specialization to actually create real "business solutions" that are more specific than a generic toolkit. So, if you like ACS/pg, and you want to start working for a company that is 100% committed to the open-source process in its development *and* in its services model, you should get in touch with me.

One final note: our commitment, and more specifically *my* commitment to ACS/pg remains unchanged. My availability will vary, but I will continue to hack and coordinate as much as I can. There is plenty of good stuff in the works for ACS/pg, and our dev community is growing very successfully. So let's keep this up! Open-Source is the only way to make a real difference.....

Collapse
Posted by Guan Yang on
Note that I don't know anything about PostgreSQL or PLS, but the PLS Callable Personal Librarian (CPL) is a C interface to the PLS engine, recommended by Philip Greenspun as the best search engine technology ever, much better, he argues, than Oracle interMedia. It might be easy to integrate it with PostgreSQL?

(Of course, he has turned into a big Oracle fan lately, so he may tell you something quite different today.)

Collapse
Posted by Ben Adida on
Yes, PLS is definitely a candidate, but first AOL needs to fully open-source it and we need a Linux port :)
Collapse
Posted by Guan Yang on
Then are the PostgreSQL and ht://Dig licenses compatible?