converting cvs to bzr branches

[Alexander Belchenko]

Reinhard Tartler пишет:
> I've been googling a bit around
about converting a cvs checkout into a
> bzr branch. My results were a bit confusing. Some recommend using
cscvs
> to convert to gnu/arch, and then from there to bzr, and some are
> recommending setting up tailor to do this job.
>
> I don't have access to the CVS Repository itself, I only have
access to
> checkout it. I'd also like to continously import changes from the
cvs
> tree (actually, I need to import 3 cvs modules, which I'd like to
have
> in 3 different bzr branches).
>
> What do you guys recommend nowadays?

I try another way: convert cvs repo to svn repo (with cvs2svn tool)
and then convert svn dump to bzr repo with svn2bzr tool. Because my cvs
repo consist many branches this way give me better result than tailor.
But not ideal, mostly because when my cvsnt-based repo converting to
svn it lost merge-points info. Mostly because cvs2svn does not designed
for dealing with this information I guess.

I try tailor but I don't understand how it should work and how to force
them dealing with branches. I don't like tailor -- it too complex to me
and has weird UI. And tailor works much slower than pair
cvs2svn+svn2bzr, maybe because I run them on cygwin (tailor does not
like cvsnt and don't want to work on pure windows+cvsnt).

BTW, cvs2svn written in python (and tailor too) and I think it possible
to extend it to dealing with merge-points info and even to make cvs2bzr
tool. But I don't have inspiration for this task.

--
Alexander