Confused about stability of revno's

[Stefan Haller]

Stefan Haller <lists at haller-berlin.de> wrote:

> Seriously though, the workflow would be to use "bzr ci" if it lets you,
> and "bzr ci --local" if it doesn't; then you would have to "bzr up" at
> some point to merge your changes back (which somehow feels backwards,
> but maybe I could get used to it).

I'm still having a hard time getting my head around the concepts here.
It seems that this:

  bzr checkout master Bob
  cd Bob
  ... hack hack hack ...
  bzr ci --local
  (Alice pushed to master meanwhile)
  ... hack hack hack ...
  bzr ci --local
  bzr up

does something entirely different than this:

  bzr branch master Bob
  cd Bob
  ... hack hack hack ...
  bzr ci
  (Alice pushed to master meanwhile)
  ... hack hack hack ...
  bzr ci
  bzr merge

I think I would prefer something closer to the second way, because it
just feels more natural.

In Mercurial, for example, when Bob tried to push after doing his local
commits, he would be told that he can't because he diverged from master;
so he would pull; and the pull works, it just creates a new head.  Bob
then merges the two heads together (locally) and pushes, and after that
all clones have the same history again.  This seems like a much simpler
concept to me.

I'd be glad if you guys could convince me that bazaar is equally simple
in concept and I'm just not getting it yet; because in all other
respects bazaar is so much superior to hg.


-- 
Stefan Haller
Berlin, Germany
http://www.haller-berlin.de/