Git and Large Binaries: A Proposed Solution - current situation?

From: Christoph Buchner <bilderbuchi@phononoia.at>
Date: Jan 8, 2013, 22:51 UTC
Message-ID: <50ECA2F8.3030703@phononoia.at>

Hello git devs!

We are currently trying to deal with the large-binaries-in-git problem, and I found this conversation from 2011 on this mailing list: http://git.661346.n2.nabble.com/Fwd-Git-and-Large-Binaries-A-Proposed-Solution-td5948908.html

I was also motivated by finding this git GSoC 2012 proposal:

Show 6 quoted lines

> Git copies and stores every object from a remote repository when 
> cloning. For large objects, this can consume a lot of bandwidth and 
> disk space, especially for older versions of large objects which are 
> unlikely to be accessed. Git could learn a new alternate repository 
> format where these seldom-used objects are stored on a remote server 
> and only accessed on demand.

What both this proposal and the email discussion proposed (among others), i.e. storing large binaries outside of git, and especially fetching them from somewhere else only on demand, sounds like it would solve our problem pretty well.

My question (I asked first in the git-devel IRC channel) is if there was any more activity on this, and/or if this is on a roadmap or similar? sparse clone sounds like something similar, but this has been pushed far back, right?

I am aware of external mechanisms (e.g. git-annex, git-media), but we would prefer something git-internal: Our userbase is heavily cross-platform (including windows), but there's no windows support for git-annex (which otherwise sounds like we could use it).

thank you for any input, all the best, Christoph