[Moderation] Welcome to the first 20 subscribers!

[David Earl]

On 01/10/2009 07:35, John Smith wrote:
> Someone else mentioned something about the best way to deal with
> potential conflicts due to later changes was to approach this on a
> node by node and way by way and relation by relation basis rather than
> attempting to make one big revert, this could be still achieved via a
> single changeset though.

Absolutely. A revert is merely an automated edit, not actually backing 
out of a previous change. When that isn't possible because of conflicts, 
or simply because (perhaps a small) part of the changeset isn't 
erroneous we need to be able to know what the conflicts were, what 
failed, and to be able to revert the bulk of it while tackling the 
exceptions individually.

David