Re: Question about cvs2svn and CVSNT

Hi Michael,

Thank you very much for your response.

I am not a programmer and don't really have the expertise required to review
and compare source code between CVS and CVSNT.

I may end up writing some shell scripts that will allow me to pick
significant points in time from a source CVSNT repository and easily port
them over to a new svn repository (i.e. a script to check out from HEAD
of branch A from CVSNT repo A and check into HEAD of /branches/branchA in
svn repo A).
Rather than focusing on keeping all the history from CVSNT, it may be good
enough for me to simply identify significant points in time from the source
CVSNT repository and check them into the appropriate target branches / tags
directories in the target svn repo.

Again, I really appreciate you answering my e-mail.

-Thank You






On Mon, Jun 7, 2010 at 11:58 PM, Michael Haggerty <[email protected]>wrote:

> jason kendall wrote:
> > I noticed on the FAQ page for cvs2svn
> > (http://cvs2svn.tigris.org/faq.html#cvsnt) that converting CVSNT
> > repositories to svn using cvs2svn may be problematic and is not
> supported.
> >
> > Does anyone know if any patches of cvs2svn are planned that might
> > address this?
>
> It would be great to support CVSNT conversions better, to help CVSNT
> users get out of their quasi-proprietary world and onto a modern VCS.
> But I know of no current work in this direction.  I think that CVSNT
> support is waiting for volunteers or sponsor$.  Maybe you would like to
> play one of those roles?
>
> A big chunk of the work of adding robust CVSNT support is to figure out
> exactly how CVSNT repositories differ from standard CVS repositories.
> That will take some work poring through CVSNT documentation,
> repositories, and perhaps even source code.  We already know about some
> of the differences but probably not all of them.
>
> Then a bit of work would have to be done on the RCS parser that is used
> by cvs2svn to retain "newphrases" that are used by CVSNT but which the
> parser currently discards.  The main point here is to handle the "kopt"
> value, which specifies, revision by revision, what expansion options to
> use for the file contents (text vs. binary, keyword expansion, etc).
> (Standard CVS only allows such options filewise.)
>
> It might also be nice to handle the commitids that are used by CVSNT and
> by newer versions of CVS, though I consider this support to be optional
> because our changeset detection is already pretty reliable.
>
> Let me know if you would like to tackle any of these tasks and I will
> give you some tips to help get you started.
>
> Michael
>

------------------------------------------------------
http://cvs2svn.tigris.org/ds/viewMessage.do?dsForumId=1670&dsMessageId=2618595

To unsubscribe from this discussion, e-mail: [[email protected]].