[Bucardo-general] Starting SWAP replication

[Rosser Schwarz]

On Fri, Jan 22, 2010 at 4:58 AM, Michal Bicz
<michal.bicz at iloopmobile.com> wrote:
> Is this the normal behaviour or should it look at both tables and mirror
> them automatically (by settting onetimecopy)?

Someone please correct me if I'm wrong, but I think the issue of
trying to use onetimecopy  with a swap sync is one of determining
which is the source and which is the destination: which do you copy
from, and which do you copy to?  In a pushdelta sync, that's clear; in
a swap, which is essentially a bi-directional pushdelta, not so much
-- particularly when both sides are populated.  If one side were
empty, it would be fairly clear, too (or at least clearer).  If not,
however, the best it can do is guess.  What happens if it guesses
wrong?

Looking briefly at Bucardo.pm (version 4.4.0), the block from lines
2765-2772, which handles onetimecopy situations, only executes if the
sync is a pushdelta.  That supports my theory, above, so I'd say that
you're seeing the intended behavior.

rls

-- 
:wq