On 2015-08-27 19:19:35 +0100, Dean Rasheed wrote:
> It also seems to me that this warning has proved its worth
Same here - I plan to re-submit it. Perhaps the number of bugs it found
convinces Tom, after I address some of his points.
> although I don't think it's something a production build should be
> producing. Perhaps it could be an Assert?
It's currently protected by a #ifdef USE_ASSERT_CHECKING. A warning
seems to make it easier to actually run the whole regression test, and
it's consistent with what we do in a bunch of other places.
Greetings,
Andres Freund