Jacques Le Roux wrote: > http://www.wandisco.com/php/pr.php?prdate=2010-02-17
Not one to complain, but that page is so lacking in details, even the linked pages are bare of anything substantial. Offline commits sound cool, and fixing the svn poop problem(scattered .svn folders all over) is nice, there are lots and lots of other things in git that are useful. Having the entire previous history immediately available, is immensly useful. When someone reports a problem, I can use git bisect to find the problem. I don't have to wait for svn to download the deltas between all old versions. Same is true when I'm trying to just look at previous code. Being able to have a checkout of trunk, with all my current uncommited work, and any staged commits that haven't been pushed upstream, then suddenly having to stop all work, to fix some critical bug in a deployed version, is very nice. I do a git stash(takes all modified files and saves them temporarily). Then, I do a git co $old_production_release_branch. Edit, commit, build deployment. git co trunk, git stash apply, and continue where I left off. When we finally decide to upgrade $old_production_release, I check it out, then do the normal git svn rebase command. The last time I did this, it only took me half a day to upgrade a 6 month old version. That is vastly superior to the old way we did it(by hand). It was made simpler this time around, because I forced myself to put all changes/fixes upstream first, before adding them to the old package.