On 11-07-11 10:12 AM, Christopher Browne wrote:
> On Mon, Jul 11, 2011 at 9:26 AM, Steve Singer<ssinger at ca.afilias.info>  wrote:
>> On 11-07-08 06:57 PM, Christopher Browne wrote:
>>>
>>> I have a would-be merge-in of the latest bug#218 changes under way;
>>> hopefully will complete that Monday.
>>>
>>> I ran thru the regression tests, and had a pretty mixed result, as
>>> follows:
>>>
>>> ->    % cat testResult.test.txt
>>> 1,17,test1
>>> 9,0,testdatestyles
>>> 7,20,testddl
>>
>> testddl can fail due to EXECUTE SCRIPT race conditions. I remember you
>> talking about an WIP patch to that issue for 2.2.  (this is putting the DDL
>> in sl_log)
>
> Ah, yes.  Bug #137.
> http://bugs.slony.info/bugzilla/show_bug.cgi?id=137
> https://github.com/cbbrowne/slony1-engine/tree/bug137
>
> I haven't touched it in a while - seems proper to wait until we're
> done with 2.1 release.
>
> The one piece of the merge that has left me uncertain is that at one
> point, MOVE SET was locking sl_event_lock, but some versions are
> locking sl_config_lock instead.
>
> I *think* it ought to just be sl_config_lock, right?

I checked in a change related to that last week.

MOVE_SET must lock sl_event_lock because the moveset_int stored function 
creates a ACCEPT_SET event.