On 2015/05/25 9:16, Peter Geoghegan wrote:
> AddForeignUpdateTargets() actually won't be called with ON CONFLICT DO
> UPDATE, and so it isn't exactly true that the only obstacle to making
> FDWs support ON CONFLICT DO UPDATE is around inference of arbiter
> unique indexes on the foreign side. It's *almost* true, though.
I think that those are interesting problems. Wouldn't we need some
additional hacks for the core or FDW to perform an operation that is
equivalent to dynamically switching the ExecInsert/ExecForeignInsert
processing to the ExecUpdate/ExecForeignUpdate processing in case of a
conflict?
Best regards,
Etsuro Fujita