Re: Performance of large queries

[prev] [thread] [next] [Date index for 2004/07/15]

From: Tony Bowden
Subject: Re: Performance of large queries
Date: 23:05 on 15 Jul 2004
On Thu, Jul 15, 2004 at 10:05:20PM +0100, colm-cdbi@xxxxxxxxx.xx.xx wrote:
> the current search doesn't seem to
> lend itself to you being able to write a wrapper to it that doesn't
> involve objects if you'd like a middle ground between objects and
> roll-your-own SQL.  Just a thought.

There are plans afoot for splitting out a lot of the search stuff to
separate methods. I don't quite know what they'll all be yet, but when
people need to override search in any way, it's not fine-grained enough
atm.

There are really 4 main processes: The bit that processes and validates
the search arguments, the part that changes these into SQL, the part
that executes that SQL, and the part that turns the results into
objects.

Ideally each part of this should be replacable with something different.

Adding a 'raw' results method should become much easier after that.

I'm not that convinced about adding it into Class::DBI itself, but
making it trivial to add to a subclass (ideally a one liner) has merit.

Tony

Performance of large queries
cdbi-talk 07:00 on 15 Jul 2004

Re: Performance of large queries
colm-cdbi 09:27 on 15 Jul 2004

Re: Performance of large queries
Tony Bowden 09:35 on 15 Jul 2004

Re: Performance of large queries
colm-cdbi 09:58 on 15 Jul 2004

Re: Performance of large queries
Tony Bowden 10:04 on 15 Jul 2004

Re: Performance of large queries
Perrin Harkins 17:13 on 15 Jul 2004

Re: Performance of large queries
cdbi-talk 19:22 on 15 Jul 2004

Re: Performance of large queries
colm-cdbi 21:05 on 15 Jul 2004

Re: Performance of large queries
Tony Bowden 23:05 on 15 Jul 2004

Generated at 11:34 on 01 Dec 2004 by mariachi v0.52