selectively applying changesets

Seth Falcon sethfalcon at
Thu Sep 29 02:52:34 CDT 2005

I'm finding the discussion of selective changeset application
interesting and have a couple naive questions:

On 28 Sep 2005, mpaesold at wrote:
> hg clone main feature1 # clone main repo for new feature
> hack ... hack ...  in feature1/, commit as needed
> After some time, you recognize that you have gone a wrong way, but
> the initial work is good, you figure out which revision is still
> "good", than do: hg clone -r a4d830edc742 feature1
> feature1-next-attempt and go on with the feature1-next repo. You can
> delete feature1 or keep it for reference.
> Finally, when you think you're done you can push to the main repo.
> For each feature you create your own branch.
> This avoids having "bad" changesets in the main repo at all.

How is such a feature different from generating a patch to apply to
the main repo instead of doing a pull?  

That is, hack on feature repo with many commits and reverts lots of
unwanted history, then just make a patch to apply to the main repo and
throw away the feature branch.

The above described interface sounds convenient, but is there
something more that I'm missing?


+ seth

More information about the Mercurial mailing list