"Karsten Hilbert" <Karsten.Hilbert@gmx.net> writes:
>> And maybe having one huge index managing the uniqueness across partitioned
>> data just defeats the idea of data partitioning!
> Except when you want uniqueness across all partitions.
Well, the point was that if the partitioning arrangement guarantees to
put distinct ranges of the key into distinct tables, then a separate
unique constraint on each table would suffice to guarantee global
uniqueness.
You can set up such a thing today, but it's a manual jury-rigged affair.
An automatic partitioning system would be a lot nicer.
regards, tom lane