3533 Tor Didriksen 2012-01-27
WL#5957 post-push fix: dont count_bits unless we are going to use them
modified:
sql/handler.cc
3532 Guilhem Bichot 2012-01-26
Final review comments for
Bug#13596330 - EXTRA ROW ON SELECT WITH NESTED IN CLAUSE
+ IS NULL WHEN SEMIJOIN + BNL IS ON
@ mysql-test/include/join_cache.inc
make the test engine-agnostic (I verified that it would still
fail without the code fix)
modified:
mysql-test/include/join_cache.inc
mysql-test/r/join_cache_bka.result
mysql-test/r/join_cache_bka_nixbnl.result
mysql-test/r/join_cache_bkaunique.result
mysql-test/r/join_cache_bnl.result
mysql-test/r/join_cache_nojb.result
sql/sql_join_buffer.cc
sql/table.h
=== modified file 'sql/handler.cc'
--- a/sql/handler.cc 2012-01-26 13:09:59 +0000
+++ b/sql/handler.cc 2012-01-27 11:15:59 +0000
@@ -4604,12 +4604,13 @@ handler::multi_range_read_info_const(uin
because the number of rows with this value are likely to be
very different than the values in the index statistics.
*/
- const int keyparts_used= my_count_bits(range.start_key.keypart_map);
+ int keyparts_used= 0;
if ((range.range_flag & UNIQUE_RANGE) && // 1)
!(range.range_flag & NULL_RANGE))
rows= 1; /* there can be at most one row */
else if ((range.range_flag & EQ_RANGE) && // 2a)
(range.range_flag & USE_INDEX_STATISTICS) && // 2b)
+ (keyparts_used= my_count_bits(range.start_key.keypart_map)) &&
table->key_info[keyno].rec_per_key[keyparts_used-1] && // 2c)
!(range.range_flag & NULL_RANGE))
rows= table->key_info[keyno].rec_per_key[keyparts_used-1];
No bundle (reason: useless for push emails).
| Thread |
|---|
| • bzr push into mysql-trunk branch (tor.didriksen:3532 to 3533) WL#5957 | Tor Didriksen | 30 Jan |