Re: Adding fulldisjunctions to the contrib

On 8/26/06, Andrew Dunstan <andrew(at)dunslane(dot)net> wrote:
> this is inaccurate, irresponsible and insulting to those of us who spend
> time maintaining pgfoundry.

Andrew,

I'm sorry if it sounded that way... it wasn't meant as such.

> It is not a graveyard. Plenty of stuff outside the core gets included in
> packaged distributions - just see for example what goes into the Windows
> distro, or the packages that CP distributes.

I'm not saying that *everything* on pgfoundry is junk... but I can
start naming dead projects if you'd like. It's like SourceForge
before SourceForge jumped the shark... now 90% of SourceForge is
either projects dead-and-gone or which hadn't even started. It's
almost not even worth the time to search SF.net anymore. I believe
that's the direction pgfoundry is headed. Not because of poor
management or administration... just that when you have a large number
of projects, the majority of which are dead or not even worth viewing,
it takes the credibility of the site down as a whole. Look at
gborg... there was some good stuff there and there still is; if you
already know about it. Both gborg and pgfoundry have projects on
there won't even work with a current version of PostgreSQL.

Outside of all us hackers... how many people actually use pgfoundry?
Does anyone have the stats? Has anyone polled users? How many of the
users are newbies and how many are already familiar with PostgreSQL?
If we don't have these basic answers, continuing to praise pgfoundry
as the home for all-things-PostgreSQL is pointless.

> The implication of your statement is that anything not accepted into the
> core is automatically somehow considered unworthy.

Not at all. I'm referring to this case in particular.

> Please refer to Tom's recent remarks about playing on extensibility
> as one of our strengths.

I never said it wasn't... extensibility is, IMHO, our *core* strength.
However, I don't think that's a good reason for pushing everything to
pgfoundry.

> My impression (please correct me if I'm wrong) is that proper full
> disjunction support would include grammar support, in which case contrib
> is not where it should belong anyway. If that's so, then the next step
> would be for somebody to pick up the work that Tzahi has done and take
> it the rest of the way. That would be a worth goal for 8.3.

You are correct, a *full* implementation would most likely include
integration into the core; grammar and all. However, being as it's an
entirely new feature in any database system ever seen, I don't think
it should be required. It's kind of funny though; it's difficult
enough to convince -hackers to adopt a feature that every other
database system in the world has, yet we're going to make it even
more difficult for an innovative feature. I can only imagine trying
to get a consensus on the grammar and implementation of a totally
nonstandard feature that only a few people really understand.

As I see it, the full disjunction code will likely end up being a low
profile project on pgfoundry because Tzahi won't have time to continue
maintaining it and not many of us have enough insight into it to do so
ourselves. As such, I don't think it's going to get enough attention
and enough of a user following to make it worth the time of one of the
core developers to pick it up.

Of course, I may always be wrong. Perhaps pgfoundry is more popular
than I've seen in past experience. Maybe one of the core developers
does want to pick up full disjunctions for 8.3.

Guess we'll just have to wait and see...

--
Jonah H. Harris, Software Architect | phone: 732.331.1300
EnterpriseDB Corporation | fax: 732.331.1301
33 Wood Ave S, 2nd Floor | jharris(at)enterprisedb(dot)com
Iselin, New Jersey 08830 | http://www.enterprisedb.com/