On Tue, Dec 5, 2017 at 4:11 AM, Alvaro Herrera <alvherre@alvh.no-ip.org> wrote:
> Tom Lane wrote:
>> Yeah, agreed. The only thing I'm concerned about back-patching is
>> the places where a wrong errno might be reported.
>
> If we're currently reporting "could not open dir: Success" then
> backpatching such a fix is definitely an improvement. OTOH if currently
> we have opendir() trying to report a failure, then LWLockRelease replace
> the errno because something completely unrelated also failed, having the
> message report exactly the opendir() failure rather than the lwlock
> failure is surely also an improvement.
Note I am +/-0 with exposing ReadDirExtended in back-branches, because
there is no use for it yet there. Only fixing the actual bugs with
errno is of course fine for me if my initial message was not clear for
back-branches.
--
Michael