Andres Freund <andres(at)anarazel(dot)de> writes:
> Either way, I'm not convinced that handling query cancels in isolation
> is really the right thing. I think pretty much all forms of interrupt
> would need to be processed, not just cancels.
+1
regards, tom lane