On Wed, Aug 07, 2019 at 10:17:25AM -0400, Tom Lane wrote:
> Not objecting to the patch, exactly, just feeling like there's
> more here than meets the eye. Not quite sure if it's worth
> investigating closer, or what we'd even need to do to do so.
Yes, something's weird here. I'd think that the index only scan
ensures a proper ordering in this case, so it could be possible that a
different plan got selected here? That would mean that the plan
selected would not be an index-only scan or an index scan. So perhaps
that was a bitmap scan?
> BTW, I realize from looking at the plan that LIKE is interpreting the
> underscores as wildcards. Maybe it's worth s/_/\_/ while you're
Right. Looking around there are much more tests which have the same
problem. This could become a problem if other tests running in
parallel use relation names with the same pattern, which is not a
issue as of HEAD, so I'd rather just back-patch the ORDER BY part of
it (temp.sql is the only test missing that). What do you think about
the attached?
--
Michael