Re: Re: Worg pull broken?

[tycho garen <garen@tychoish.com>]

On Wed, Oct 27, 2010 at 09:47:24AM +0200, Sébastien Vauban wrote:
> - Does using ssh give you compression for free?  Hence, a quicker way to
>   download big repos?

Git can interact with the remote repository differently, allowing it
to do some de-duplication over SSH that it can't do over HTTP. Also,
depending on your SSH configurations SSH can (and generally is, I
think, correct me if I'm wrong) somewhat compressed (But then, so is
HTTP with DEFLATE/gzip sometimes). This might not speed up the
operation both because of the processor overhead to do the
compression, but also because SSH's encryption takes a bit of
overhead. 

Basically, "it depends on what free is," and pragmatically git pull's
over SSH are quicker than git pulls over HTTP. 

> - Can you use whatever protocol (git, http, git+ssh) in front of any "git
>   URL"?  I mean, if I see on the Web, http://xxx.git, can I replace http by
>   git with no impact?

No. SSH (and git) protocols specify the path of the file based on
where the file lives on the host machines file system. Start at the
root / (or chroot) in the top level and go from there. The paths for
HTTP are dependent upon the configuration of the web server and are
often somewhat distinct from what is actually going on in the file
system given virtual hosting and URL rewriting. 

Hope this helps. 

Cheers,
tycho

-- 
tycho(ish) @
 garen@tychoish.com
  http://www.tychoish.com/
  http://www.cyborginstitute.com/
  "don't get it right, get it written" -- james thurber