On Mon, Mar 10, 2008 at 11:00 AM, Justin <justin@emproshunts.com> wrote:
>
> That comment was not meant to be an insult or disparaging in any way what
> so ever. If it was taken as such then i'm sorry.
I am sure it would have been fine in person, I just think over email
it sounded abrasive.
But could you stop topquoting please?
> It seems the biggest performance hit is copying of the array content from
> one memory variable to another which is happening allot.
Yeah, I think arrays just can't handle a whole lot of data, that is
all. They are "tricky", and shouldn't be used for heavy lifting (more
than 1k of elements feels like you are asking for trouble).
> I'm not really against using a temp tables to hold onto values. I used to
> do that in Foxpro when i hit the hard limit on its array but other problems
> start popping up. If we use a temp table keeping track what going with
> other users can make life fun.
I think temp tables have scope, though you should test this, so that
you can use them with impunity in functions and not worry with
multiple users.
> I really want to figure this out how to speed it up. I have to write allot
> more aggregate functions to analyze R&D data which will happen latter this
> year. right now this function will be used in calculating manufacturing
> cost.
I think a combination of aggregate functions along with some more
design would be best. For example: can you have a trigger calculate
the normalized weight of a row on insert? Have triggers keep another
table with summary information updated as you modify the data? Etc.
There is a lot to PG that would help for this kind of thing.