[darcs-users] [OT] Larry McVoy on the Bitkeeper licence

[David Roundy]

On Tue, Feb 15, 2005 at 03:08:58PM +0100, Karel Gardas wrote:
> On Tue, 15 Feb 2005, David Roundy wrote:
> > Right, but he seems to think that even providing the graph of the linux
> > kernel repository would give away too much information about how BK
> > works to competing developers.  It doesn't seem to be possible, for
> > example, to develop a script that would convert directly from the BK
> > kernel repository to a darcs repository using darcs.  Debugging such a
> > script would reveal too much of BK's internal workings--at least that
> > was what Larry has hinted.
> >
> > Of course, he's always *also* said that one could legally use BK itself
> > to get data out of BK, but he just hasn't explained how one could
> > legally do this.  I imagine one would have to do it without contacting
> > me, lest one be tainted as a contributor to darcs.
> 
> I don't know if I'm not completely wrong here, i.e. misunderstanding you,
> but IIRC Larry always stressed the idea of rsyncing BK repository and
> using GNU SCCS (http://cssc.sourceforge.net/index.shtml) to get sources
> from it, which is 100% legal way and you can keep developing darcs.

True, but then you'd need to reverse-engineer BK's data format without the
benefit of BK.  Plus, one gets the joy of learning to use a revision
control system that refers to CVS as "a more modern replacement"...  I've
never wanted to look at BK's internal data format, but don't imagine it'd
be pleasant.  And I've never heard of anyone actually using cssc to examine
it.  But you're right that that would be the absolutely safe legal way to
go about mirroring the kernel repository without half the changesets
missing.
-- 
David Roundy
http://civet.berkeley.edu/droundy/