* Li-Bing.Song@stripped <Li-Bing.Song@stripped> [10/08/13 13:10]:
> 3479 Li-Bing.Song@stripped 2010-08-13
> Bug #39804 Failing CREATE ... SELECT doesn't appear in binary log
> Bug #51501 RBR, slave stops with HA_ERR_KEY_NOT_FOUND
First, resolving 47899 is a pre-requisite.
You can't attempt to fix an edge case in RBR replication of CTS
unless you settled your mind on how you replicate the main case.
Consequently, this bug should not be fixed in 5.1, but in 5.5 and
on.
Second, I think we shouldn't attempt to support this.
We are better off disallowing stored functions that modify
tables in DDL. Completely. I don't see how it can be useful
unless the intent is malicious.
If we decide to support this, we need a worklog and additional
QA testing. Let's not try to fix it in scope of WL#5370.
Thanks,
--
kostja
