As per numerous reports this morning, PG 8.2.2 and 8.1.7 both fail on
fairly simple scenarios involving typmod-bearing columns (varchar,
numeric, etc) with check constraints or functional indexes (and maybe
other cases too, but those are the ones reported so far). I have not
been able to reproduce the failures in 8.0 but I think it may have the
same issue in a weaker form. We need a quick re-release I'm afraid.
I have applied a patch that resolves the problem AFAICT, but this time
around it would be nice to get some more eyeballs and testing on it.
Please try CVS HEAD or branch tips this afternoon, if you can. Core
is currently thinking of wrapping update tarballs this evening (maybe
around midnight UTC?).
Also, if anyone can find a related failure in 8.0.11 please notify us.
I'm not sure whether we need an update in that branch or not ...
regards, tom lane