Re: SSH URI draft

[Joe Salowey <jsalowey%cisco.com@localhost>]

The original intention was for interactive SSH shell sessions.   I agree that port forwarding should be out of scope for this.  I also tend to think that other subsystems would be out of scope as well.   I hadn't really thought about shell sessions without pty, did you have something in mind?  

The draft needs to be clear about this so I'll work up some text.   Do you see a problem with using "ssh" as the scheme name to mean  SSH shell sessions with pty?


On Nov 17, 2010, at 12:18 PM, Nicolas Williams wrote:

> On Wed, Nov 17, 2010 at 10:54:54AM -0800, Joe Salowey wrote:
>> I submitted a new version of the SSH URI draft from several years
>> back.  I removed the SFTP and SCP schemes since I could nto find a
>> stable reference.  We can address these protocols in a separate
>> document.   Please review and send comments.  I'm particularly
>> interested to know what URI schemes have been implemented for these
>> protocols.  
>> 
>> http://tools.ietf.org/html/draft-salowey-secsh-uri-00
> 
> Nothing about what resources can be referred via SSH URIs other than the
> entire SSHv2 service?
> 
> Resources I can think of:
> 
> - shell sessions, with and without pty
> - port forwardings
> - subsystems
>    - paths to resources for those subsystems
> 
> Now, to be fair, the idea of specifying port forwardings (and agent, X11
> display, and other forwardings) in URIs scares me, so I'd rather leave
> them out than in.  And clearly SFTP merits its own URI scheme, and
> perhaps so do all subsystems.  But ISTM that we should at least consider
> distinguishing between "the whole service" and "shell session".
> 
> Nico
> --