Re: pgsql: Switch TAP tests of pg_rewind to use a role with minimal permiss

On Sat, Apr 13, 2019 at 11:12:25AM -0400, Alvaro Herrera wrote:
> I'm not sure that reverting and re-committing a slightly improved patch
> is a better approach than just waiting for the (presumably small) fix to
> be committed. Leaving a small number of animals in red for a small
> period is not that bad.

Thanks, however I am not so sure when it comes to Windows failures as
these can prove to need tricky treatments. Anyway, there is a daily
commit activity and it is always annoying to bump on failures from
others and this could hide other problems. 947a350 and 9daefff are
for example two things from Noah which refer to jacana and have een
committed in-between, so reverting temporarily was still the best
thing to do I think.

I have committed the fixed version of this patch, after making sure
that the test is able to work on Windows, and that this does not
impact what we want to test: if one of the GRANT queries is removed
the test fails properly.
--
Michael