On Sat, Jul 19, 2008 at 05:11:43PM -0700, Tracy R Reed wrote:
One of the nice things about having a centralized VCS is that if the HD in my current developer workstation bites the dust I have uploaded all of my commits to the centralized VCS. If I'm using git or hg and the changes are kept local until I push them I could lose a lot of work.
This argument is fairly easy to turn around into making the opposite argument. With 'git' there is no reason for me to keep my changes locally. It is very easy to create alternate branches on another repo, or even just setup a backup repo. Pushing to these repos is very fast once most of the history is there. For the most part, the DVCS is much better from a backup point of view, since _everybody_ has all of the history. For the centralized VCS, the single repo becomes a critical resource. If it fails, everything stops. We had some developers where permission problems prevented them from being able to get to our main git repo. I pushed the repo to a network drive they had access to, and had a script update this push once a minute. It consumes very little resources, and suddenly they have access again. David -- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
