On Wed, May 4, 2016 at 5:02 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Very good point, but unless I'm missing something, that is not what the
> current patch does. I'm not sure offhand whether that's an important
> estimation failure mode currently, or if it is whether it would be
> sensible to try to implement that rule entirely separately from the "at
> most one" aspect, or if it isn't sensible, whether that's a sufficiently
> strong reason to confine the "at most one" logic to working only with FKs
> and not with bare unique indexes.
Tomas seems to feel that that is what the current patch does, and
indeed that it's the main point of the current patch, but you seem to
think that it doesn't do that. Either I'm misinterpreting what one of
you is saying, or you are missing something, or his patch fails to
accomplish its intended purpose. It seems important to figure out
which of those things is true.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company