When will be CVS replaced by modern version control system?

[Karel Zak]

On Thu, Nov 08, 2007 at 11:08:19AM +0100, Adam Tkac wrote:
> On Wed, Nov 07, 2007 at 11:41:58AM -0600, Josh Boyer wrote:
> > On Wed, 7 Nov 2007 15:12:45 +0100
> > Adam Tkac <atkac at redhat.com> wrote:
> > 
> > Replacing a VCS for the fun of it is pretty pointless.  Can you
> > elaborate on a workflow you would like to see that CVS is not suited
> > to?  Right now, CVS works fine for what we do, which is mostly editing
> > spec files.

 Nonsense.

 My work is searching and fixing bugs, rebasing, reviewing, testing
 and backporting patches. The spec files are decoration around this
 work... Sorry, but editing spec files is 0.001% of my work on
 Fedora/RHEL.

> > I am by no means a proponent of CVS.  I think it sucks.  But we have no
> > _usecase_ for a different VCS at the moment.
> > 
> 
> It's not replacement for fun. Yes, CVS works and I believe it will
> work to end of universe. But question is if We have something better
> than CVS. And We have. There're some common problems (yes, CVS and
> SVN suffer :) )

 Few notes:

 * many people around Fedora are still not well educated about new
   content tracking tools. So we are not ready for the change.

 * many people still think about VCS as about a patches/source code
   archive -- that's very wrong. A good content tracker is a
   __development tool__. For example with GIT you can do non-linear
   development, rebasing, prototyping, bug bsearch, easily
   send/receive patches by mail, generate customized changelogs,
   scripting, off-line work, ...

 * __unfortunately__, we don't maintain source code in our VCS! We use
   it for *.patch files + commit messages. It means you can't use all
   modern features -- just because GIT, Hg, ... are designed for work
   with source code (unlike Quilt, StGIT).


    Karel

-- 
 Karel Zak  <kzak at redhat.com>