[postgis-devel] Re: [postgis-users] GIST index speed

[Tom Lane]

I wrote:
> I wonder if it'd be sane for the TOAST code to keep a small lookaside
> cache of recently-detoasted values ...

I put up a design proposal for that on pgsql-hackers:
http://archives.postgresql.org/pgsql-hackers/2008-06/msg00709.php
It seems do-able as new development for 8.4, but of course it'd be
much too invasive to be considered for back-patching.  I don't currently
have any ideas about a back-patchable fix that would alleviate the
performance problem for index filter conditions.  If anyone's got bright
ideas, please contribute to the -hackers thread.

My previous patch might be worth pursuing in any case, on the grounds
that 50% improvement is better than nothing ...

			regards, tom lane