SVN-like merging (contrib)
Sébastien Pierre
sebastien at xprima.com
Thu May 11 12:17:35 CDT 2006
Le jeudi 11 mai 2006 à 13:05 -0400, Colin McMillen a écrit :
> > > If this is the case, then "parent/base/other" would be better.
> >
> > No, that confuses me even more.
> >
> > local = current = mine = the file in the working directory
> > base = common ancestor of local and other
> > other = remote = what should be the new version if I discard my changes
>
> I think it's a bad idea to use "local" or "remote" in a distributed SCM,
> as it's very likely that "base" and "other" could also be "local" files
> (in the sense of existing on the same machine, not necessarily being in
> the same repository). So "current" is more clear. So
> "current/base/other" makes the most sense to me, though "parent" instead
> of "base" would also be intuitive and acceptable (which leaves us back
> at Sébastien's suggestion).
Hmm... so the question is : is the current nomenclature (as found in the
hgmerge) correct ?
I agree that "local" used with "remote" is confusing, because remote
could come from a local repo, even the current repo.
However, "local" without remote is fine, because we usually talk about
"local" changes.
Now, "parent" would be innappropriate if the given file is the common
ancestor. In this case "base" or "ancestor" are both fine (ancestor
being more natural).
With this in mind, the "base/local/other" or "ancestor/local/other"
would be two appropriate choices. The first one being preferred because
it is consistent with the current nomenclature.
Is that OK for everybody ?
-- Sébastien
More information about the Mercurial
mailing list