Joseph Galbraith wrote:
Do we need to bump the protocol version? I'm pretty sure we are already shipping something that say s it conforms to version 2 that will squawk at the new protocol number. I think there may be others.
On the one side, it's pretty easy to bump the revision. On the other side, Internet-Drafts blah blah...reference material blah blah...work in progress.
My guess would be, most implementations would see the new status code and spit it at the user as being unknown, possibly ending the (publickey) connection. Also, this is only an issue if the status code gets sent at all - if you're only sending attributes returned by listattributes, it shouldn't...
If this does turn out to be a problem, the new status code could die and we define GENERAL_FAILURE as the code to send then - but that seems like a poor compromise.
-- Jon Bright Silicon Circus Ltd. http://www.siliconcircus.com