Teodor Sigaev <teodor@xxxxxxxxx> writes: > GiST may work with any subset of index columns too. Even in existing code I > don't see any problem except NULL in a first column. GiST doesn't store tuples > with leading NULL value (gist.c lines 174, 326), so index doesn't contained them. Well, that's exactly the problem :-(. Or at least one of the problems; the other being what you'd use as search key to find such tuples. > After our work about WAL-lization GiST, it may work with "invalid" > tuples (possibly occured after crash recovery), so itsn't a big deal > to add support NULL in a first column. But freeze date is > outdated... Should I add or leave it to 8.2? Too late for 8.1 I'd say --- this definitely sounds like a new feature rather than a bug fix. regards, tom lane ---------------------------(end of broadcast)--------------------------- TIP 6: explain analyze is your friend