On N, 2005-05-12 at 16:56 -0400, Jan Wieck wrote:
> On 5/12/2005 12:53 AM, Sven Willenberger wrote:
> > 
> > Jan Wieck presumably uttered the following on 05/11/05 19:24:
> >> On 5/11/2005 5:29 PM, sven at dmv.com wrote:
> >> 
> >>> Comment Added to bug
> >> 
> >> 
> >> Question about your plperl stuff:
> >> 
> >> Is it that the updates are not recorded in sl_log_1 at all, or is it 
> >> that the slon just doesn't immediately generate a SYNC event?
> >> 
> >> 
> >> Jan
> >> 
> > 
> >  From what I can see a) the first set of rows was never replicated and 
> > currently sl_log_1 is empty which leads me to the conclusion that it was 
> > never recorded. Per the suggestion in the bug thread I will try a couple 
> > other ideas to help pinpoint the issue.
> 
> Are you saying that using a plperl function you can reproducably manage 
> to get your origin and subscribers to have different content while at 
> the same time sl_log_1 is empty?
> 
> If that is the case, I would be highly interested in a self contained 
> test case.

I think that this depends on the order of firing the triggers. - if the
trigger fires after slony's trigger and updates the same row, then it
seems quite logical that replication of later changes is not noticed by
slony.

-- 
Hannu Krosing <hannu at skype.net>