"Joachim Wieland" <joe@mcknight.de> writes:
> Zdenek Kotala sent in regression tests for SET a few days ago which got
> turned down. I think however that the idea has merit and that only his
> implementation was not useful. Attached is another regression test script
> that executes some SET / SET LOCAL within transactions and subtransactions.
Applied with minor editorialization --- mostly, avoiding an unnecessary
dependency on the installation configuration.
regards, tom lane