On 11/02/2015 01:27 AM, Patrick Lauer wrote: >>> >> I wonder what would happen if we just... dropped a full clone of >> gentoo.git onto the rsync mirrors? >> >> > You find out how many users have limited bandwidth (again) > > Why on earth would you do the most wasteful hybrid strategy when people > are in general trying to optimize for efficiency? >
I was only jokingly pointing out that we don't have to switch away from rsync to update a git repo on the users' machines. And it's OBVIOUSLY stupid... but is it really? Limited bandwidth wasn't the problem (stated in this thread) with using git, an unreliable connection was. Using rsync to transfer the git tree takes care of that, even if its wholly inefficient. Once users have the full git repo on their machines, they have two options. They can update it efficiently with `git pull`, or they can update it with rsync by using `emerge --sync`. You can even mix the two, and the repo that we're pointing at could have a copy of the metadata pregenerated like the github mirror does. The full git log would be there, and it doesn't require us to overhaul any of the mirror infrastructure. I'm not actually suggesting this so nobody has to talk me out of it.
