From: Ilari Liusvaara <ilari.liusvaara@elisanet.fi>
Date: Wed, 02 Dec 2009 05:50:48 GMT
Subject: Re: [RFC PATCH 0/8] Git remote helpers to implement smart transports.
Message-ID: <20091202055048.GA31244@Knoppix>
In-Reply-To: <20091201193009.GM21299@spearce.org>

On Tue, Dec 01, 2009 at 11:30:09AM -0800, Shawn O. Pearce wrote:
> Ilari Liusvaara <ilari.liusvaara@elisanet.fi> wrote:
> > 
> > For instance, to support new types of authentication for smart transports
> > without patching client git binaries (SSH has lots of failure modes that
> > are quite nasty to debug) or abusing GIT_PROXY (yuck). 
> 
> So the bulk of this series is about making a proxy for git://
> easier to tie into git?

This is making the "layer 5/6" parts of git:// easier to replace, for whatever
reason that replacement may be desired (and the lower layer is just assumed to
be some kind of full-duplex link).

The part about abusing GIT_PROXY is _very_ nasty hack to be able to layer 6
gateway git smart transports.

The git:// protocol stack is: 
- Git smart transport subprotocols (upload-pack, upload-archive and receive-pack)
- git:// (request signaling and data passing)
- TCP/IP (or comparable)

And ssh://:

- Git smart transport subprotocols (upload-pack, upload-archive and receive-pack)
- SSH (request signaling, data passing, encrypt & auth).
- TCP/IP (or comparable)

Smart-HTTP:

- RPC versions of git smart transport subprotocols
- HTTP
- TLS (optional)
- TCP/IP (or comparable)

This is about:

- Git smart transport subprotocols (upload-pack, upload-archive and receive-pack)
- Some prtocol layer(s) (request signaling, data passing, maybe encrypt & auth, etc...)
- TCP/IP (or comparable)

> Forgive me if I sound stupid, but for gits:// shouldn't that just
> be a matter of git_connect() forking a git-remote-gits process
> linked against openssl?  Or, maybe it just runs `openssl s_client`?

gits:// was just an example. There can be other interesting stuff (I don't
even pretend my imagination is the limit). And I would rather link the gits://
handler to GnuTLS than OpenSSL, but that's seperate matter...

As for "other interesting stuff": Smart transport using Kerberos auth (just
throwing ideas, probably not going to implement that)?

> Why go through all of this effort of making a really generic proxy
> protocol system when the long-term plan is to just ship native
> gits:// support as part of git-core?

gits:// is not actual goal of this series. Its just something to build on
top of it.

-Ilari