Bruce Momjian <pgman@candle.pha.pa.us> writes:
> Well, pg_group is not used very much. Can we disable the cache from
> using the index? I think that would get us out of the problem.
But they're unique indexes ... can you guarantee that a unique index
that's only getting told about some of the updates to its table is
not going to kick out any inappropriate errors? It might be OK but
it doesn't give me a warm feeling.
I think we were kinda stuck on this one. The correct fix is clear
and I don't think we'd have been doing anyone any favors by trying
to invent a workaround. I'm just glad we found it now and not a
few days *after* 7.0 release...
regards, tom lane