> > Well, how many people want to vote for Andreas' suggestion
> of having
> > both
> >
> > int pg_cancel_backend(int)
> > bool pg_backend_cancel(int)
> >
> > with the former deprecated but still there for backward
> compatibility?
>
> Oh no, what have I started!! :-)
>
> Let's just make the change and let the few people affected
> modify their scripts, otherwise this is gonna get very messy.
Yeah. As one who has been bitten when I tested a system on the new
version, I still think it's reasonably easy. It's not like queries using
pg_cancel_backend are scattered throughout an app. In my case it's in a
single place, and it's easy enough to work around it.
(My solution: write a tiny wrapper SQL function that looks different
when running on 8.0 and 8.1)
If we deprecate it and say "gone in 8.2", I'm going to have to write
this wrapper anyway. And if we don't set a schedule for when it's gone,
we might as well not deprecate it, and that would make it very messy...
> Thankfully I think we've all learnt from this :-)
Yup, that's the main thing. This time it wasn't too bad (as it only
affected a seldom-used function), it might not be next time if we do it
again.
//Magnus