[PATCH] alter merge command to only consider active branch heads by default

Benjamin Pollack benjamin at bitquabit.com
Tue Mar 17 21:31:15 CDT 2009


On 3/17/09 7:46 PM, Matt Mackall wrote:
>
> This breaks tests quite badly. Which suggests we need to rethink the
> approach here and in fetch.
>
> 'Inactive' is not really a good cue not to use a branch. Users regularly
> make important branches temporarily inactive by starting a new branch at
> their heads.
>
>   

I agree with both parts, except that, in the case of fetch, whose point
is to make Mercurial more accessible to new users, I think that trying
to merge only active heads by default--which is what the patch I sent
this morning does--is probably fine.  You can still explicitly fetch
inactive branches by manually specifying revisions.  I think that
behavior is fine.

The above patch I sent more because I wanted to figure out whether we're
doing the right thing at UI and API levels with inactive branches.  You
definitely sometimes wish to merge into an inactive branch; the question
is whether inactive branches should be consulted by default.  For
implicit merges, I don't think they should be.  For querying the state
of the repository, I think they should.  At the moment, we're doing both.

--Benjamin


More information about the Mercurial-devel mailing list