On Thu, Jul 10, 2014 at 04:38:54PM -0400, Greg Troxel wrote:
> So it sounds like we need to do a few things.  It's certainly fair that
> you have not enough time to hack on Coda, but currently no one else
> (outside CMU) can commit, not even longstanding well-known members of
> the extended coda community.
> 
> If you are not wanting to work in CVS, that's a (valid) clue that the
> time has come to move from CVS to git.   So I guess there's the question
> of converting with history, using reposurgeon or something.  Then,
> there's how to host the git repo so that specific other people are
> allowed to push to it.

The problem with having the canonical repository in CVS is that it
doesn't allow for forks to live. Also the repository is stored in AFS,
so to give anyone direct access is hard because they need to have an AFS
client and the right setup so that CMU's server can accept their
kerberos credentials and stuff I haven't even begun trying to figure
out.

Most of our newer projects are already using Github, we actually have a
organizational account with quite a few public and private repos (no
Coda repos yet though). It also helps with tracking pull requests/code
reviews and such.

> My understanding is that Rune has fixed the major ick if not all of it
> with clog, and has something far more useful than the current code.  So
> it should go in the repo, to be usable in the interim until the glorious
> future arrives.  (IMHO, setting up a separate ad-hoc Needham-Schroeder
> system with custom code that hasn't had the wide review and fixes that
> Kerberos has instead of just relying on Kerberos seems totally
> unreasonable in 2014; I get it that afs/coda coevolved with krb from
> long long ago and there are historical reasons.)

The modular clog still uses the ad-hoc Needham-Schroeder system with
custom code that hasn't had the wide review etc. because that is how a
Coda client and a Coda server set up their RPC2 connections. It only
changes the password authentication step where the Coda token is
obtained from the auth2 daemon before clog hands it to the Coda client.

> So it seems like we need
> 
>   decide to move to git
> 
>   find a volunteeer to create reposurgeon config files or whatever to
>   effect a repeatable scripted conversion from CVS to git
> 
>   decide the conversion is good enough
> 
>   convert and decommission CVS
> 
> and then there is the external write, merging of various things,
> fixing/redoing the release scripts.

That sounds like a good approach to me. I have to looks at this
reposurgeon, I think I just used cvsps in the past to turn the CVS
history in a whole lot of patches that then got applied to create the
Coda git repo.

Jan