Bruce Momjian <bruce@momjian.us> writes:
> Tom Lane wrote:
>> Shouldn't that be back-patched?
> Uh, well, it is going to change the behavior of back branches, and
> because we only got one report of the bug which has existed since 8.2, I
> didn't want to risk it. Should I?
I would say that the odds of the initial BEGIN failing are negligible
anyway, so what it boils down to is whether a failure on the final
COMMIT needs to be reported. Seems to me the answer is "yes", and the
only reason we haven't had more complaints is that not too many people
have actually relied on the exit status. Anyone who *does* look at the
exit status is not going to be happy with the current behavior.
In short: it's a bug, fix it.
regards, tom lane