Hi,
sorry, some strange key-combination made my mail client send too early...
I myself wrote:
> I was trying to say that Postgres-R internally relies only on a unique
> index with not null constraint. It doesn't care if you name it PRIMARY
> KEY or REPLICATION KEY or whatever.
>
> So, it's just a question of the syntax. We already have PRIMARY KEYs,
> and those are pretty much what I think is needed in 99% of all cases as
> the pointer to the replication
.. as the pointer to the index to use for replication.
Offering the user a possibility to choose another (index + not null)
would require something like ALTER TABLE ... ADD REPLICATION KEY ... or
some such. Mostly syntactic sugar, which can be added as soon as we
really need it.
> While I'm normally an absolute fan of generality,
>
> I think you didn't quite get the point.
Iiik.. that's what I didn't want to send and wanted to delete before
sending... :-) Sorry.
Regards
Markus
