Tom Lane wrote:
> Alvaro Herrera <alvherre@2ndquadrant.com> writes:
> > Stephen Frost wrote:
> >> I'm guessing the right answer here is to just add test_pg_dump to that
> >> list.
>
> > I don't like this solution, because it means pg_dump will get little
> > testing on Windows.
>
> No, that's incorrect: @contrib_excludes stops the MSVC stuff from trying
> to *build* in that directory, which is fine because there's nothing to
> build. It doesn't prevent running the test case. See eg.
>
http://buildfarm.postgresql.org/cgi-bin/show_stage_log.pl?nm=thrips&dt=2016-05-06%2015%3A37%3A27&stg=testmodules-install-check-C
> where the test_extensions test does get run even though it's in
> @contrib_excludes.
Ah, that's surprising, but much better of course. Works for me.
> In any case, reverting the patch would mean no Windows testing, so
> your argument doesn't seem to have much force even if it were true?
I was suggesting to revert it for the beta and put it back with a
working Windows implementation after the beta, before GA. I was afraid
that a hack to hide the test from Windows would become permanent.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services