"0 conflicts encountered."

[Martin Pool]

On  2 Dec 2005, Jan Hudec <bulb at ucw.cz> wrote:
> On Fri, Dec 02, 2005 at 22:14:46 +0100, Matthieu Moy wrote:
> > Aaron Bentley <aaron.bentley at utoronto.ca> writes:
> > 
> > >> And when are we thinking to switch to --merge-type=weave?
> > >
> > > I'm not sure.  I actually prefer three-way diffs, most of the time
> > > - they give you a BASE file, so you can see what changes OTHER made.
> > 
> > AAUI, --merge-type does a 3-way merge at the file granularity (file
> > names and meta-information), and a weave merge for file content. So,
> > it shouldn't be difficult to put a file.BASE in the tree in case of
> > conflict, right? (but I guess getting the ancestor in the conflict
> > markers would be meaningless)
> 
> I am not really sure about it, but shouldn't the weave merge be equivalent to
> 3-way merge if the base is well defined*? In that case, the logic could be:
> Do 3-way merge if the ancestor is well defined, weave-merge otherwise.
> 
> * 3-way merge base is well defined if there is exactly one most recent common
>   ancestor.

Yes, it should be at least very close to that behaviour.  The other
difference would be that (as I recall) 3-way merge currently determines
a single common ancestor for the whole tree, whereas weave merge
determines the virtual ancestor per-file.  

I think we could fairly easily as the weave code to produce a BASE
version as well: either the mushed-together virtual ancestor, or one
ancestor text.  If there's a clear ancestor, they'll be the same thing.

-- 
Martin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20051206/4e556c8a/attachment.pgp