Re: efficient compare

Bob Badour wrote:

> But if you consider the example I gave previously that involved a single
> insert into each table, swapping logs would be trivial.
>
> Likewise, by comparing the size of the log to the size of the relation
> variable, one could choose to send the relvar instead of the logs.
> However, one would then have trouble reconciling multiple changes to
> similar tuples.

I want a generic algorithm. You are right that swapping logs is excellent for your example, but what if one node does not have half of what the other node has, then logs might be a bad idea. I want to periodically run this algorithm to synchronize the nodes. Something similar to what I described in the other posting (the algorithm I gave), but maybe something more efficient. Received on Mon Apr 24 2006 - 10:38:19 CDT