[CMake] CMake 2.8.0 RC 1 ready for testing!

[Michael Wild]

On 26.09.2009, at 19:33, Eric Noulard <eric.noulard at gmail.com> wrote:

> 2009/9/26 Michael Wild <themiwi at gmail.com>:
>> Usually I use
>>
>> cvs update -dARP
>>
>> which IMHO should probably be the default, because that's what  
>> people (i.e.
>> me) expect from an update.
>
> -dRP are ok even if I think R is not needed, however you usually don't
> want A unless you want to checkout cvs HEAD.

As I said, people refers mainly to me, and that's the behaviour I  
want... But you're right, of course.

>
>> Are there any plans for CMake (and other Kitware projects) to  
>> switch over to
>> a more contemporary VCS, such as git, hg or bzr (in the order of my
>> preference :-))?
>
> You may already set up a git cvsimport repository which may be
> updated daily/hourly. That way you may git clone + branch
> from the cvs-imported repo and use git instead of cvs.
>
> If the cvs imported repo is publicly available
> other may benefits from your GIT repo.
>

I am doing this with ParaView, but I find it to be rather fragile.  
Especially the option which supposedly should prevent "noisy  
commits" (can't remember it's name right now) caused git cvsimport to  
actually lose commits. Strange thing is, the commits showed up in the  
log of the CVS parsing program just fine... I think the main problem  
is that CVS only tracks individual files and has no notion of commit  
sets.

Michael