Tom,
> I never said that they couldn't develop useful views. I do question
> the assumption that they can be all things to all people. I think the
> claims being made in this thread are highly overblown. In particular
> I doubt that views that expose everything anyone might want to know
> are going to be amazingly much simpler than the underlying catalogs.
Well, we're actually trying to answer 2 seperate questions on this thread:
1) would some kind of new system views in theory be valuable and accepted, and
2) what needs to change in these particular system views.
We need to answer question (1) first because there's no point on doing a lot
of work to modify our proposal if it's just going to be rejected.
> So far the goals that have been presented include being complete, being
> simple, and being stabler than the underlying catalogs. I think you
> can have *one* of those; probably not two and definitely not all three.
> So tell us the truth about what your priorities are and which goals will
> lose when there's a conflict.
My personal goal is to have views which are complete and simpler than the
system catalogs to the extent that it doesn't interfere with being complete.
For super-simple, incomplete views we have the information_schema and the old
system views. I think you've sucessfully punched several holes in the "more
stable" argument, so that's really the lowest priority.
> I see a whole lot of NIH (Not Invented Here) syndrome --- the developers
> of these views love 'em, sure, because they fit the way those developers
> think. But that doesn't prove everyone else will love 'em.
Sure, but so far the arguments have not revolved around the specific views
proposed, but rather the whole concept of new system views at all. In fact,
I think Peter was the only person to give us any useful feedback on the
actual design ...
--
--Josh
Josh Berkus
Aglio Database Solutions
San Francisco