On Thu, Mar 11, 2010 at 8:32 AM, Niels Reedijk <niels.reedijk@xxxxxxxxx> wrote: > > The next step is documenting a workflow for developers with the > alternative SCM. There are two lines of doing that; the first would be > along the line of getting a personal migration of the repository with > hgsvn or git-svn. That is the only way to get a git or hg repository > that can exchange data with the svn repository. (As far as I found > there is no way to share the metadata that the svn converters use to > keep track of the status of svn and migrated repositories). I believe it is possible to edit the git-svn information in a git repo which was originally cloned through git-svn. I at least intend to try this with the Haiku git repo I cloned from Travis' repo at git.newos.org. The main issue is just using the right information when pushing changes back upstream. If I get this working I will document it so others can do it as well, because cloning the svn repo and all the commits into git (and probably hg too) is a long and slow process (plus it puts a load on the svn server, which could potentially impact other committers.) I can also try to develop a decent work flow for the git-svn bridge and then document that as well. > For non-svn committers we can design another type of documentation; > one that shows how hackers can use the cloned hg or git repository and > use that to track patches that they (and others) made. I don't know about Mercurial, but I know Git was designed to ease the management and applying of patches, so it might even make sense to advise people to use one of these alternate workflows if they are not svn-committers. -- Regards, Ryan