Code review comment for lp:~tsarev/percona-server/5.1_bug_838725_fix

** after your changes have_innodb_plugin.inc is functionally identical to have_innodb.inc. Which means there's no need for s/have_innodb/have_innodb_plugin/ in percona tests
** mtr is just a symlink to mysql-test-run.pl, there's no need to duplicate diff changes
** please revert the "Running test cases" changes. they might be good for debugging, but they don't carry any functional changes, so they can only increase maintenance cost.
** MySQL bug #62340 reported by you is invalid. The difference in results has been introduced by your changes in Percona Server, it has nothing to do with upstream.
** main.index_merge_innodb, main.rowid_order_innodb, main.type_bit_innodb and main.endspace result differences look fishy. I understand that there may be a difference in results between builtin InnoDb and plugin/XtraDB. But those tests used to pass with XtraDB in older Percona Server releases. Isn't it something we should look into instead of just updating .result files? In other words, what's changed in those tests or XtraDB between Percona Server 5.1.55 and 5.1.58 that leads to results being different?
** I would like a link to a Jenkins build on all slaves, not just the centos ones.