Brian,
As Andy and I responded to Elwyn, this was discussed within the working
group. The consensus was at the time that the client knows its a client
and the server knows its a server so this isn't really a problem. I was
in the minority at the time. My preference would be two separate BEEP
profiles, and this was in fact the case in an earlier version of the
document, as I recall.
Eliot
ps: whatever tool you're using to generate this email is generating
illegal destination addresses in the "To" field.
Brian Carpenter wrote:
Discuss:
(Based on Gen-ART review by Elwyn Davies)
The base NETCONF protocol spec talks about 'client/server' instead of 'manager/agent'. This document uses 'manager/agent' but needs to be
consistent with the base spec.
The base NETCONF protocol spec requires (in section 2):
The transport protocol MUST provide a mechanism to indicate the
session type (client or server) to the NETCONF protocol layer.
So where does the BEEP mapping do this? If that isn't defined,
it's a bug.