Thread: pgsql: As long as we're forcing an ORDER BY in these four join queries,
pgsql: As long as we're forcing an ORDER BY in these four join queries,
From
tgl@svr1.postgresql.org (Tom Lane)
Date:
Log Message: ----------- As long as we're forcing an ORDER BY in these four join queries, we had better make sure the sort order is totally specified; else we get burnt by platform-specific behavior of qsort() with equal keys. Per buildfarm results. Modified Files: -------------- pgsql/src/test/regress/expected: join.out (r1.21 -> r1.22) (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/test/regress/expected/join.out.diff?r1=1.21&r2=1.22) pgsql/src/test/regress/sql: join.sql (r1.14 -> r1.15) (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/test/regress/sql/join.sql.diff?r1=1.14&r2=1.15)
Just as an FYI ... this commit made it into RC1 ... anything after this won't unless I re-tag/package ... On Fri, 3 Dec 2004, Tom Lane wrote: > Log Message: > ----------- > As long as we're forcing an ORDER BY in these four join queries, we had > better make sure the sort order is totally specified; else we get burnt > by platform-specific behavior of qsort() with equal keys. Per buildfarm > results. > > Modified Files: > -------------- > pgsql/src/test/regress/expected: > join.out (r1.21 -> r1.22) > (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/test/regress/expected/join.out.diff?r1=1.21&r2=1.22) > pgsql/src/test/regress/sql: > join.sql (r1.14 -> r1.15) > (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/test/regress/sql/join.sql.diff?r1=1.14&r2=1.15) > > ---------------------------(end of broadcast)--------------------------- > TIP 4: Don't 'kill -9' the postmaster > ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
"Marc G. Fournier" <scrappy@postgresql.org> writes: > Just as an FYI ... this commit made it into RC1 ... anything after this > won't unless I re-tag/package ... That's fine. I'm going to look at Kris Shannon's assert failure on trying to cast one composite type to another, but I don't feel a compelling need to have the fix in RC1. ("There's always another bug.") regards, tom lane