pgsql: Note case where nbtree VACUUM finishes splits.

2019-03-04 Thread Peter Geoghegan
Note case where nbtree VACUUM finishes splits. The nbtree README claims that VACUUM can never finish interrupted page splits by design. That isn't entirely accurate, though. Note an exception to the general rule. Discussion:

pgsql: Disable dump_connstr test on Msys2

2019-03-04 Thread Andrew Dunstan
Disable dump_connstr test on Msys2 For some reason the dump test with names with high bits set fails on Msys2 (although not Msys1). Disable the tests for now, so that other tests can run. Branch -- master Details ---

pgsql: Allow recovery tests to run on Windows as an admin user

2019-03-04 Thread Andrew Dunstan
Allow recovery tests to run on Windows as an admin user This is the only test that fails when run as an admin user. The reason is that when Postgres is started via pg_ctl its admin privileges are lowered. However, this test called 'postgres -D datadir' directly, resulting in a failure. Replace

pgsql: Correct obsolete nbtree page split WAL comment.

2019-03-04 Thread Peter Geoghegan
Correct obsolete nbtree page split WAL comment. Commit 2c03216d831, which revamped the WAL record format, failed to update a comment referencing the old API. Update the comment. Branch -- master Details ---

pgsql: Reorder configure tests for accept() in Windows

2019-03-04 Thread Andrew Dunstan
Reorder configure tests for accept() in Windows Currently only frogmouth in the buildfarm uses the 32bit params, and it's not able to build past release 10, so put those last, saving substantial configure time on more modern systems. Even if we get a modern 32 bit Windows system at some stage we

pgsql: pg_partition_ancestors

2019-03-04 Thread Alvaro Herrera
pg_partition_ancestors Adds another introspection feature for partitioning, necessary for further psql patches. Reviewed-by: Michaƫl Paquier Discussion: https://postgr.es/m/20190226222757.GA31622@alvherre.pgsql Branch -- master Details ---

pgsql: Test partition functions with legacy inheritance children, too

2019-03-04 Thread Alvaro Herrera
Test partition functions with legacy inheritance children, too It's worth immortalizing this behavior, per discussion. Discussion: https://postgr.es/m/20190228193203.GA26151@alvherre.pgsql Branch -- master Details ---

pgsql: Remove volatile from latch API

2019-03-04 Thread Peter Eisentraut
Remove volatile from latch API This was no longer useful since the latch functions use memory barriers already, which are also compiler barriers, and volatile does not help with cross-process access. Discussion: