Try to stabilize flappy test result. This recently-added test case checks the plan of an inner join between two identical tables. It's just chance which join order the planner will pick, and in the presence of any variation in the underlying statistics, the displayed plan might change. Add a WHERE condition to break the cost symmetry and hopefully stabilize matters.
(We're still trying to understand exactly why the underlying statistics aren't as stable as intended, but this seems like a good change anyway, since this test would surely bite us again in future.) While here, clean up assorted comment spelling, grammar, and whitespace problems. Discussion: https://postgr.es/m/4168116.1711720...@sss.pgh.pa.us Branch ------ master Details ------- https://git.postgresql.org/pg/commitdiff/c2df2ed90a82fc0acd07ed88ac433350edfd6b54 Modified Files -------------- src/test/regress/expected/subselect.out | 14 ++++++++------ src/test/regress/sql/subselect.sql | 13 +++++++------ 2 files changed, 15 insertions(+), 12 deletions(-)
