On Mon, Nov 30, 2009 at 12:42:29PM +0100, Stefan Sperling wrote:
> On Thu, Nov 26, 2009 at 08:01:30PM +0100, Daniel Näslund wrote:
> > cp-add
> > -------------
> > alpha has been locally copied to eta, incoming add eta
> > Options:
> >   THEIRS:1) svn revert eta
> >          2) rm eta
> >          3) svn up eta
> 
> It would help to change the wording of the spec to be more precise.
> You could describe what Subversion should be doing at the level of
> manipulating meta data of nodes in the working copy, rather than
> listing svn commands. See the wc-ng design notes for suitable
> terminology (e.g. BASE tree, WORKING tree, etc.).

Yeah, I will do that. This was an initial attempt at verifying how
conflict resolving in practice would work. I will continue with the
conflicts when doing a merge. Hopefully some patterns for organising the
resolver i starting to show by then.
 
> What do you mean by "svn up eta"?
> 
> The resolver will be called during a running update.
> The BASE of eta needs to be updated to the new revision no matter what,
> otherwise a tree conflict can put you in a cycle that prevents you from
> updating forever.

That's something that I do need to remember!
 
> Do you mean something like "replace eta in WORKING with the new BASE
> that came in during the update"?

Yes, that is exactly what I mean. I plowed through the WC-NG docs last
summer and understood... Not much of the states in the versioned tree.
This time around WORKING, ACTUAL and BASE might actually mean something!

Thanks for your feedback,
Daniel

Reply via email to