--- In firebird-support@yahoogroups.com, Lester Caine <lester@l...>
wrote:

> ... It creates a difference file between the source and the
> destination and copies ( compressed if selected ) the diff which
> it then applies to the destination copy. My understanding of
> is a little more hazy, but I think that it will maintain BLOB
> data in the same order between 'runs' so that rsync should not
> have to coy them.

I have been reading the tec-doc at
http://samba.anu.edu.au/rsync/tech_report/, and because of the rolling
checksum the order of the data should not matter. It builds af hash
table much like zip compression. If you use rsync with a blocksize of
x bytes then the source file can have block A before B and the
destination can have B before A. The rsync algorithme would still
catch it and only transfer if source A block <> destination A block.
The position of the block should not matter. That's the magic of
rsync!

> SINCE you have already run GBAK, and have an 'off line' file
> I can't see that it a problem. It WOULD be a problem if trying
> to do it on the live database. I think Jaume missed that?

That is what I thought too.

Brian Vraamark