Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
>> Don't you ever run the tests before checking stuff in?
> ... Nope. I do it once a batch is checked in. I assume submitters have
> already done it. Too much work.
Whether or not you assume that the submitter ran regress tests, it's
unsafe to assume that a several-days-old patch is still good against
current CVS tip. Especially not when you get merge failures, as you
certainly must have in this case.
regards, tom lane