On Fri, Jul 11, 2014 at 10:27:50AM -0400, Jan Harkes wrote:
> On Thu, Jul 10, 2014 at 06:44:01PM -0400, Adam Goode wrote:
> > The most reliable way I have found for CVS transitions is cvs2svn followed
> > by git-svn. Both tools are quite good at what they do.
> 
> I tried cvs-fast-export and reposurgeon last night, it got something out
> but I haven't dug too deeply into the resulting repository. Anything
> after 2007 looks pretty good, things get worse as you go further back.

Actually the earliest commits in the CVS history make very little sense.
it looks like anything before the following commit,

    4.0 checkin         Peter Braam     1996-11-22 14:48:02

is not very useful, there are some BSD kernel code and utils-src/mond
(Coda server resource? monitor daemon) but nothing else. Only by the
time it hits that 4.0 checkin all the bits and pieces have been pulled
together.

Also noticeable is that quite a few commits can/should be merged. For
instance there is a series of "maria's changes 11/18/96" which is spread
over about 15 individual commits. That series also makes it clear that
CVS only tracks committer and not author. In the early days a reference
to the actual author is typically made in the commit message, for the
later parts where I used git in parallel to CVS I am going to have to
pull author information out of the corresponding git commits.

> > Can you make the raw CVS repo available somewhere? I might have some time
> > to play with this.
> 
> Yes, I'll make some archives of the Coda, LWP, RPC2 and RVM
> repositories. An interesting bit is that LWP, RPC and RVM originally
> were part of the Coda tree, but at some point got split off into their
> own repositories, I believe by copying the underlying RCS from one CVS
> repo to the new ones.

Total size of all CVS repositories is only 11MB compressed. I've put it
at http://coda.cs.cmu.edu/~jaharkes/coda-mirror-20140711.tar.gz.

I also started a github project to help with the conversion. Already
added a bootstrap script that downloads and unpacks the CVS tarball
and fixes up some of the tags that cvs-fast-export tripped up on, and a
file to map author / committer names to their corresponding full names.

    https://github.com/cmusatyalab/coda-git-conversion

Jan