----- Original Message -----
Hi Daniel!
The test case should also look at the result of the:
select * from mysql.index_stats where index_name='a' and table_name='t1';
Added in the test case.
Thanks
Also, I got the explanation slightly wrong. It's not index_init that causes the problem, it's calling ha_index_first with a garbage buffer as a parameter that messes things up. The flags in table specifically say that (in that case) the table->record[0] is garbage, but we used it anyway.
Good to know there isn't a lower level lurking bug because of this. Thanks for the explanation.
diff --git a/mysql-test/t/mdev-7362.test b/mysql-test/t/mdev-7362.test new file mode 100644 index 0000000..6551893 --- /dev/null +++ b/mysql-test/t/mdev-7362.test
more recent test cases include the area of the test failure in the filename or include the test with the general t/statistics.test
Hmm, I made a new file to test for this bug and named it statistics_index_crash.test. I don't think it fits going into the statistics.test.
Good description. like it.
Can you please add a test case for innodb too as it supports FT in 10.0.
Done.
thanks
If this is the fix the if..return can go much earlier in collect_statistics_for_index, before Index_prefix_calc() even.
Done.
thanks -- -- Daniel Black, Engineer @ Open Query (http://openquery.com.au) Remote expertise & maintenance for MySQL/MariaDB server environments.