>> descriptions. However, I believe pg_dump is currently in a
>> broken state with respect to the dumping of indexes,
>> possibly related to the current work done with
>> INDEX_MAX_KEYS. Can anyone confirm this?
> I just looked and I don't see any problems.
It was broken --- I committed fixes for it at about 11PM EST.
(The same problem as for func args --- it wasn't coping with
zero suppression in oidvectorout/int2vectorout.)
regards, tom lane