On Mon, 2008-10-20 at 18:45 -0400, Tom Lane wrote:
> Simon Riggs <simon@2ndQuadrant.com> writes:
> > On Mon, 2008-10-20 at 17:44 -0300, Alvaro Herrera wrote:
> >> That's been "extended with an epoch counter" per the docs; I don't think
> >> that's appropriate for the new functions, is it?
>
> > I assumed it was, so you can subtract them easily.
>
> > It can be done either way, I guess. Happy to provide what people need. I
> > just dreamed up a few that sounded useful.
>
> I don't think you should be inventing new functions without clear
> use-cases in mind. Depending on what the use is, I could see either the
> xid or the txid definition as being *required*.
The use case for the two functions was clearly stated as "together
allows easy arithmetic on xid difference between master and
slave". In that context, xid plus epoch is appropriate.
There are other use cases. We can have both, neither or just one,
depending upon what people think. What would you want "xid only" for? Do
you think that should replace the txid one?
This is everybody's opportunity to say what we need.
> In any case, do not use the wrong return type for the definition you're
> implementing.
err...Why would anyone do that?
--
Simon Riggs www.2ndQuadrant.com
PostgreSQL Training, Services and Support
