On 6/6/2006 12:04 PM, Christopher Browne wrote:
> Either way, it would substantially complicate the subscription process :-(.

The problem isn't the long running transaction on the subscriber, but 
the one on the provider. Using multiple transactions for that means that 
the subscriber would not get one consistent snapshot of the entire set 
but rather a different snapshot for each single table. I think the first 
ever done SYNC is horrible enough as it is with its humongous where 
clause. I don't even want to imagine how that would look like if we had 
a different xid exclusion list per table in a 50+ table schema.

Since we are now substantially speeding up the copy_set, I don't see how 
Slony is more of a problem than pg_dump.


Jan

-- 
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck at Yahoo.com #