On 08/12/10 19:02, Jan Urbański wrote:
> On 08/12/10 18:45, Tom Lane wrote:
>> The real fix in my mind is to replace GEQO search with something
>> smarter. I wonder what happened to the SA patch that was reported
>> on at PGCon.
>
> I got distracted with other things :( I'll try to plan the two queries
> with SA and see what the results are. If they're good it'll certainly
> raise my motivation on finishing up the module and proposing it.
I'm pleasantly surprised that the SA code as it stands today, setting
the equlibrium factor to 8 and temperature reduction factor to 0.4, the
query takes 1799.662 ms in total. With the default values it runs
forever, but I long discovered that defaults taken from the original
paper are not well suited for my PG implementation (I could plug my MSc
thesis here, but I'm way too shy for that). 8/0.4 are values where I got
better results than GEQO for Andres' monster-query.
Maybe it actually has some value after all... Let's see if I can
untangle myself from plpython in time to clean up that code before January.
Cheers,
Jan