### Please describe the problem.
Attempting to populate one annex from another over an SSH connection is very slow.

### What steps will reproduce the problem?
1. Create an annex on Linux
2. Clone annex to Windows system
3. Set up sshd on Linux system, access keys, etc.
4. "git annex get ."
5. Checking if an individual file exists takes a few seconds, and downloading it takes a few seconds. on a 100Mbps LAN, transfers at an average of 7Mbps.
6. By comparison operating over a working HTTP connection transfers between 20~40 Mbps.


### What version of git-annex are you using? On what operating system?
Windows 7: 4.20140627-g8a36ec5 (from the git-annex download page)

Debian Linux: 3.20120629 (from the package manager)

> The problem is that git-annex is designed to be used with ssh connection
> caching. However, I have not gotten ssh connection caching to work on 
> Windows. I did try getting this to work with IIRC, cygwin's ssh, but
> it seems that the Unix sockets it expects to use with `-o ControlMaster`
> don't work, or it's not been ported to use the Windows equivilant.
>
> It's not practical to make git-annex reuse a single ssh connection
> itself, due to the way rsync is run over ssh.
> 
> Since this is basically a bug in ssh for Windows, and not in git-annex
> per se, I am going to close this bug report. However, I do mention
> this in [[todo/windows_support]].
> 
> [[done]] --[[Joey]]

[[!meta title="No ssh connection caching support on Windows"]]