I'm currently working on how we can import our large VSS repository into
Subversion. We do not want to do a full migration as we really do not
need all the old history added (and it has proven to be extremely
difficult to do). So instead we would like to import the last few
"releases" of our code into Subversion.
Importing the oldest release is very easy. It will simply become my
"trunk" and I'll also make a copy that will basically be the tag for
that release.
I now want to "import" the next release. When finished the second
release should be what is in my "trunk". I realize this doesn't provide
a "true" history. In in VSS we renamed a file it would show up in SVN
as a completely new file and the old file being deleted. That is fine
with us as people can always go back to VSS to find out what really
happened (which will only happen rarely and only for a few months).
So in the end we would have history through the past couple of release -
but only the final files in each release. But moving forward we would
have a normal SVN repository.
However, I'm not sure of the best way to do second (and subsequent)
"imports". Does anyone have any ideas how this can be done?
The obvious solution is to get a working directory of my trunk, make the
working copy look like the second release and do a commit.
Unfortunately, this is a bit more complicated then I first envisioned.
I can't just do a file sync - because for added/deleted files I actually
need to do svn commands. We are talking thousands of files here. (Over
1700 directories.) So I'm really hoping I can find a better solution.
Maybe I can import the second release into a separate space and merge it
into the trunk? This sounds like the right way to go but I haven't
thought it through... Any way, has anyone else done this? Any
recommendations?
Ray
Received on Fri Apr 16 04:37:16 2004