Re: table partitioning pl/pgsql helpers

Hi Robert,

Il giorno 01/giu/07, alle ore 04:08, Robert Treat ha scritto:

[...]

> We I set these up for our clients, I typically seperate the
> partition creation
> piece from the data insertion piece. (Mostly as partition creation,
> especially with rules, is a table locking event, which is better
> done in a
> non-critical path). If you really must do it all in one go, you'll
> have a

I see, and now I agree with you it's better to decouple partition
creation
and data insertion.

> much better chance of accomplishing this using all triggers (and
> triggers are
> better anyway)... i think you could do it with a combination of
> rules and a
> trigger (on insert to parent, create new parition and insert into
> it and
> delete from parent) but it would certainly need testing to make
> sure you dont
> have multi-rule evaluation... course since your making a trigger
> anyway...

Even more importantly, I just discovered (trying and then reading
pgsql docs) that the rule
system is completely bypassed by the COPY FROM statement, so I think
I'll rewrite
everything using some sort of trigger-generating procedure because I
want this stuff
to work transparently (and we do lots of copy from).
Thanks for the advices,

Enrico Sirola
enrico(dot)sirola(at)gmail(dot)com