Rafal,
Ok. I see your argument. Thanks for clearing that up.
Patch approved.
>> I still don't see why it needs to run against all storage engines. All
>> of the code being exercised is above the storage engine layer and
>> therefore of no consequence. Granted there are some minor things that
>> are specific to some of the engines that occur above that layer, but
>> for backup or restore it isn't related to the quality of the backup code.
>>
>
> I don't think your premise that all happens above the storage engine
> level is true. Or, at least, we can not be sure. We are checking
> interruptions and we don't know where exactly they will be caught. For
> example, one of the points where RESTORE is stopped before interruption
> is "restore_before_sending_data" which is this place in data_backup.cc:
>
>
> DEBUG_SYNC(thd, "restore_before_sending_data");
> ret= drvr->send_data(buf);
>
> if (log.report_killed())
> goto error;
>
> Thus the interruption will be intercepted inside driver's send_data()
> method and I want to see that it is correctly handled regardless of
> which driver is used. Therefore I want to keep this test in
> backup_engines suite.
>
>> Please move the test to the backup suite.
>>
