On 2015-05-08 19:32:02 +0200, Andres Freund wrote:
> If the failure is indeed caused by checkxmin (trying to reproduce
> right now), we can just remove the updates in that subsection of the
> tests. They're not relevant.
Hm. Or easier and uglier, replace the CREATE INDEX statements with
CREATE INDEX CONCURRENTLY. There's a fair bunch of tests in that file,
and keeping them update free will be annoying.
Greetings,
Andres Freund