rtree_gist merge breaks regression

Andrew Dunstan <andrew@dunslane.net> writes:

A number of buildfarm platforms appear to be failing regression tests
from ordering differences due to the rtree_gist merge.

Fixed, thanks.

BTW, could we get the owner of echidna/herring (I assume these are the
same machine) to clean out the bogus /tmp/.s.PGSQL.65432.lock file that
has been breaking its build for a couple weeks now?

regards, tom lane

Tom Lane wrote:

Andrew Dunstan <andrew@dunslane.net> writes:

A number of buildfarm platforms appear to be failing regression tests
from ordering differences due to the rtree_gist merge.

Fixed, thanks.

BTW, could we get the owner of echidna/herring (I assume these are the
same machine) to clean out the bogus /tmp/.s.PGSQL.65432.lock file that
has been breaking its build for a couple weeks now?

I have forwarded this to him. Also note that log, history and members
pages on the buildfarm web now contain owner email addresses in (what I
hope is a) non-harvestable form.

cheers

andrew