I stand corrected. Still, it will save bandwidth being all local.
Actually that is not quite true either. While yes you might save a few bytes here or there. It is really negligible for a continuous integration build server. Because the build server will have a local repository that mirrors the server repository, you will only download changes. To understand how negligible this will be, you have to understand that GIT differs from SVN in that GIT only downloads patches for changes made, not the whole file like SVN does. So often the changes (or patches) never get very large for continuous integration environments like the one we are talking about.
For example say a new line was added to some C++ file, and that is all that was changed. In SVN, it will download the whole file, in GIT it will just download a patch for that one line. As a real world example, here is a change that was made in November by Bill.https://github.com/hMailServer/hMailSer ... 001e4e4135
to see what would actually be downloaded just add patch on the end of the URL.https://github.com/hMailServer/hMailSer ... 4135.patch
And you see exactly what is going to be downloaded for that change.
Secondly ignoring all the arguments above, from what I see about the frequency of the commits and builds that are made, this is pretty much a non starter as an argument.