On Fri, Apr 30, 2010 at 5:34 PM, Tim Visher <tim.visher@xxxxxxxxx> wrote: > As far as I know, git --daemon is explicitly not supported in Windows > environments. At least that was the case not too long ago. git daemon works perfectly fine in cygwin. I've heard that the msysgit one doesn't necessarily work, but that may have been fixed. I haven't tried using the git daemon as a service, though; I mostly use it as an ad-hoc thing to easily let me exchange branches with my co-workers before they're ready to be pushed into the "real" server, thus it doesn't matter much if git-daemon doesn't restart on boot. So I can't help with the original poster's problem. > If you're running git --daemon because of its efficiency, have you > considered the fact that [smart-http][] is almost as efficient? That > should work in any http server, a plethora of which exist for Windoze. That's about 10000 times harder than running 'git daemon' from the command line, though, if all you want is ad-hoc sharing. Your suggestion is probably a good one for setting up a "real" Windows git server; setting up a Windows ssh server should also work. We just use a Linux server at work, even for our Windows users, and life is fine. > Also, file system cloning is very efficient so if you're trying to > share a repository you could in theory share over the file system > using filesystem permissions to allow for cloning. Hmm, in my experience, git cloning over NFS/SMB is kind of crappy; using a real git server is much faster. This is because of latency: the client doesn't know which bytes to read until it reads some of the other bytes, so there's a lot of back-and-forth communication. The "real" git protocols (like git daemon, smart http, ssh server, etc) let the server make these decisions and can thus go much faster. Have fun, Avery -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html