Migrating from Clearcase to Mercurial

Simon King simon at simonking.org.uk
Sat Dec 17 17:10:59 UTC 2011


On Fri, Dec 16, 2011 at 9:26 AM, Jouni Airaksinen
<Jouni.Airaksinen at descom.fi> wrote:
> You seem to have given it thinking already. Although I doubt there would be
> much problem if you used lots of named branches. Would make things slightly
> easier and you wouldn't need to worry about tagging much. You could always
> have a seperate branch for builds, so every merge there makes a new build
> (ie. on CI server which assigns the build numbers). But of course it means
> new ways of thinking about workflow.. don't be too worried about people not
> adapting. I've seen it myself, people have been thinking like Subversion and
> after a while (and possibly little bit help) they start seeing the
> possibilities and trying them out.

Hi Jouni,

Thanks for replying. I agree that there are many workflows that we
could use, including ones that are much simpler than our existing one.
I'm hoping that once we've made the switch, we can look at simplifying
our process, but to begin with I think I'd like to keep things the
same as much as possible.

I also suspect that using a named branch for every change actually
requires more effort and discipline than the many clone approach,
since a named branch has to be created in advance. This was one of the
things I disliked in Clearcase, having to create a new branch type in
the central repository before doing any work. With Mercurial, as long
as I've already got a local clone, I can start hacking, and only need
to create the server-side clone if I actually want to put my code up
for review.

>
> You can always push to a different repository than you pulled from, so that
> could be used as well so with named branches and review clone (where all
> code goes for reviewing) as push target could work and you wouldn't have to
> create "forks" if that's enough in your internal workflow. Basically what
> you've described is how bitbucket/github works, by doing separate clones.
>

Using a single server repository for code review is an interesting
idea that I hadn't considered; I will need to think about that some
more.

Thanks again,

Simon



More information about the Mercurial mailing list