Re: Request test of this bug on MySQL 5.0.51 - Bug #37439
Gordon Burditt wrote:
>> I've submitted a bug report (Bug #37439) titled
>> "Incorrect output order from ORDER BY using 2 field index in MEMORY BTREE".
>> I'm getting incorrectly ordered records out from a MEMORY database with
>> a two-field BTREE index. This fails in MySQL 5.0.27 on both Windows 2000 and
>> Fedora Core 5, two rather dissimilar systems.
>>
>> MySQL support requests that it be tested on MySQL 5.0.51. So
>> if anyone has 5.0.51 handy, I'd appreciate a test as below.
>
> I have 5.0.51a, is that close enough? Running on FreeBSD 6.2, Intel
> architecture. Quick results: it looks like it sorts correctly.
Thanks. Mind doing an
EXPLAIN SELECT * FROM ratingqueue ORDER BY rating_state;
That will settle whether the newer version is using the
index.
I get an EXPLAIN result of
1, 'SIMPLE', 'ratingqueue', 'index', '', 'rating_state', '5', '', 1544, ''
If you get an "EXPLAIN" result with "filesort" in it, it didn't use the
index.
Can anyone else get this to fail? There are at least three
previous unconfirmed bug reports since 2003 of incorrect sorting in
cases involving multiple keys where one key is an ENUM, so there
may be a hard-to-reproduce bug in this area.
John Nagle
|