On Thu, 2 Aug 2001, Perrin Harkins wrote:
> I've had great success with "Select and cache only the row keys, fetch full
> rows as needed". We were also caching the individual records (in
> BerkeleyDB), so some pages never needed to hit Oracle at all after the
> initial query. A good way to go, if your data is not too volatile.
This is more or less how Alzabo works (but you can tweak it). When it
fetches row objects, its really just fetching primary keys that match your
query. As you request other columns they are fetched and cached. You can
also specify some columns to be fetched with the primary key and specify
that others should be considered a group (when A is fetched, get B & C
too).
Actually, I stole a lot of this from Michael Schwern's Class::DBI and got
some more ideas from your talk at ApacheCon, Perrin.
So if people are interested in implementing this, they may want to just
consider using Alazbo (or you can rip out the caching code and use that
separately if you want).
-dave
/*==================
www.urth.org
We await the New Sun
==================*/
