Getting started with bzr-svn (was Re: bzr-svn and subversion revisions)

[John Szakmeister]

On Tue, Sep 8, 2009 at 5:02 AM, Jelmer Vernooij<jelmer at samba.org> wrote:
[snip]
> You should be able to do this by having a checkout of the Subversion
> branch around and merging your feature branches into that occasionally. This
> will guarantee that you never end up trying to change the mainline history. You should also be able to merge the mainline into the feature branches regularly
> without problems.

That's part the problem. :-)  The SVN repo is physically disconnected
from us, and is active.  So, I can clone it at some point, bring it
back, do our work, and then head back to do some integration.  At that
point SVN's trunk has advanced, so I really need to rebase to bring it
cleanly into Subversion. :-(  Once I do that, then I have problems on
our end because more than one person is working on the code base...
and that makes the rebase operation rather unfriendly.  I don't think
there's a clean way to handle it, other by doing multiple merges,
pushing into a separate branch, and letting the commits get reordered
there (in the branch, not trunk).  That's not really preferable,
because then the SVNites have a hard time telling what exactly
happened to the code. :-(

-John