Strategies for push/merge problem?

[Risto Kankkunen]

Greg Lindahl <lindahl <at> pbm.com> writes:
> > The situation I'm referring to is when people want to record for posterity
> > the steps they took to implement a feature instead of the zeus athena
> > alternative.
> 
> I was suggesting pushing all the history when you do push. So it's all
> there.

Sorry, I may be a bit slow here, but where do you push the history? Into the
mainline? Then you have the problem that there exist changesets that don't pass
tests (with the well known implications to bisect etc.)

I'm thinking of having the individual incomplete steps in a branch, which serves
as documentation and it also indicates that those changesets are not "safe" to
check out or bisect etc.

The changeset grouping extension was mentioned as an alternative and I'd like to
figure out the different uses cases and pros/cons for those.

I also just read that "Remember that monotone branch memberships are determined
by certs, and branches need not be fully connected." Anyone care to comment what
that means compared to Hg branches?

Having used RCS, CVS and SVN (and been forced to use SourceSafe) and having had
to merge the branches my co-workers created, I have personally developed a habit
of linearizing my changes (hg record is nice). I'm currently contemplating
whether I should use branches more to make different feature development lines
more explicit.