Kyotaro HORIGUCHI-2 wrote
> Hello, I think this is a behavior as desinged.
It may not technically be a bug but it definitely could use some TLC from
Tom Lane.
I'm curious, but too busy to check myself, how these three queries would
perform with the index columns placed in reverse order. Then create a
summary post with the 3 queries x 2 indexes, and the plan summaries (not the
entire explain) would let someone quickly see the end result without sifting
through 3 posts and lots of explanation.
Also, can you test on 9.3 and/or 9.4? Someone knowledgable may know
outright but otherwise this may have been recently improved and, because it
is not really a bug, not back-patched. Release notes should indicate this
but testing is more accurate.
David J.
--
View this message in context:
http://postgresql.1045698.n5.nabble.com/BUG-11441-Weird-and-seems-wrong-behavior-of-partial-indexes-with-order-by-limit-tp5819400p5819658.html
Sent from the PostgreSQL - bugs mailing list archive at Nabble.com.
