What replaces csup?

pete wright nomadlogic at gmail.com
Wed Sep 19 14:56:32 UTC 2012

On Tue, Sep 18, 2012 at 5:44 AM, Stas Verberkt <legolas at legolasweb.nl> wrote:
> Jerry schreef op :
>> On Tue, 18 Sep 2012 05:00:08 -0700
>> Michael Sierchio articulated:
>>> We are really behind the curve here.  Git assumes (correctly) that
>>> disk space is inexpensive, much cheaper per byte than network
>>> bandwidth.  By the time we adopt SVN completely, every serious project
>>> I know of will have moved from subversion to git.
>> If you are going to make a sweeping change anyway, it makes no sense to
>> do it in a half–assed manned. However, it does appear that in all too
>> many instances, FreeBSD plays follow the leader rather then taking the
>> bulls by the horns and getting ahead of the curve. I am sure I'll be
>> hearing from the "baby steps" choir now. In any event, a comprehensive
>> side-by-side evaluation of the two should be done by an impartial party.
> We should not be forgetting that Git and Subversion represent two different
> workflows. The latter stands for a centralistic development cycle, and the
> former for a distributed manner. Thus, this type of choice does not really
> have to do with big or small steps and leading of following, but more about
> the production cycle you want to have.
> If we were to use a Git-like system, the releng team would (probably) be in
> control on which patches are excepted from the pool of suggested changesets
> by the community of developers. This community would be more free in the
> manner in which they experiment, and there would be a less strong
> differentiation between "committers" and other people suggesting updates. On
> the other hand, our current approach has a controlled group of committers
> and the releng team only has the additional power of setting the schedule
> and taking the snapshot that becomes the release. (Gravely simplified.)
> It is a matter of taste.


one thing worth noting is that developers have been using mercurial
for quite a bit of time now for FreeBSD development(1), to take
advantage of the distributed model of that SCM.  yet having the main
tree under CVS in the past, and SVN currently, makes sense to me.  i
feel that it results in a cleaner public tree that is easier to
navigate.  so fortunately the project has been able to take advantage
of both of of these philosophies of SCM.


(1) http://wiki.freebsd.org/LocalMercurial

pete wright

More information about the freebsd-questions mailing list