Re: Sheduling in SQL

Hannu Krosing <hannu@tm.ee> writes:
> Tom Lane wrote:
>> (some of the protocol changes in 6.4 were done to make it easier ;-)).

> I may misremember, but IIRC some older protocol (or at least libpq) 
> returned 0 as backend pid to listening client if it was notified by itself.

> Currently it returns the actual pid for any backend. Is this what you 
> changed?

That was one of the smaller items.  The bigger problem was that the
backend wouldn't forward you NOTIFY events unless you issued a constant
stream of dummy queries.

> Anyhow we need some _documented_ way to get backend pid

PQbackendPID() seems adequately documented to me ...
        regards, tom lane