Marek Lewczuk <newsy(at)lewczuk(dot)com> writes:
>> I'm testing 8.3beta and I think that there is a problem with gist/gin
>> indexes. The performance of 8.3 is very bad comparing to 8.2.
Comparing to 8.2.which-exactly?
I'm wondering if you have an example that doesn't work well with the
gist page-splitting heuristic changes that Teodor put in awhile back.
Those are all in 8.2.5 already, though you might have to REINDEX to
really notice a change if the index was first built with an older
8.2.x release.
Can you provide sample data illustrating the problem?
regards, tom lane