Re: [Netconf] xml start directive with ssh

[Andy Bierman <andy at netconfcentral.com>]

Wes Hardaker wrote:
>>>>>> On Thu, 25 Jun 2009 19:50:48 -0700, Andy Bierman <andy at netconfcentral.com> said:
> 
> AB> At the time 4742 was written, nobody in the WG was aware of how
> AB> SSH subsystems actually worked.
> 
> No, it was a deliberate decision that some systems might want to only
> start a netconf agent based on a command line flag.  Subsystems were
> understood at the time, people just wanted to be able to exercise both
> potential "start" mechanisms.
> 
> 

Well, this other 'start mechanism' was never documented,
if it was discussed in any detail.  It was understood
at the time that the 'netconf' subsystem must be specified,
when starting up ssh from the command line.

There was discussion at the time about logging into the CLI,
however that is done, and then some sort of 'switch to NETCONF'
command would be issued, and the NETCONF subsystem would
then be started somehow.

I really don't remember any rules about garbage text passed to the
NETCONF subsystem before the <hello>.  What if the <hello>
is in a string or XML comment, and not 'real'?  Is the
garbage supposed to be parsed in case there is any XML or
quoted strings in there?

I think the NETCONF subsystem should only send and receive
XML instance documents.  I would be surprised if any implementations
accepted arbitrary garbage text at any time.



Andy