Oleg Bartunov <oleg@sai.msu.su> writes:
> I recommend post your problem to -hackers mailing list. I have no idea,
> what could be a problem.
I wonder whether the problem is not windows versus non windows but
original database versus copies. If it is a GIN bug it seems quite
possible that it would depend on the order of insertion of the index
entries, which a simple dump-and-reload probably wouldn't duplicate.
If you were working from a dump it'd be easy to try creating the index
before populating the table to see if the bug can be reproduced then,
but there's no certainty that would provoke the bug.
The rest of us have not seen the dump data, so we have no hope of
doing anything with this report anyway.
regards, tom lane