Other misc things in the drafts

["Joseph Galbraith" <galb-list%vandyke.com@localhost>]

Here are a number of comments on a fresh read of the
drafts, something I haven't done for a while.

- In the userauth draft, Section 6, we refer to the
  "client username on the remote host"  Since the
  "user name" field is the username on the remote
  host (from the clients point for view) I think
  this must really mean the "username on the client"

  In any case, "remote host" is vague, since we don't
  really know what point of view we are using, client
  or server.  I would suggest this just become "username
  on the client" or perhaps "client user name"

- In the connection draft, section 4.3.1, we should note
  that the x11 authentication cookie MUST be hexadecimal
  encoded.

- In the connection draft, near section 4.7 should we add
  a "terminal-mode-change" message which could be used to
  notify the other side about mode changes (for example
  from input cooked to raw mode)

  This might allow optimizations for long lag time links
  (satellite) if the client wanted to cook lines on it's own.

  I would suggest

     byte      SSH_MSG_CHANNEL_REQUEST
     uint32    recipient_channel
     string    "terminal-mode-change"
     boolean   FALSE
     string    encoded terminal modes
     string    TERM environment variable value

  We may not want to do this -- we are trying to finish
  the drafts -- and it doesn't have a proven usefulness
  (though I've had occasion to wish I could do this as we
  tried to work around NT's lack of real PTYs.)

- In the connection draft, section 5.1, we should specify
  text like this:

  The response to the "tcpip-forward" request will look like:

     byte SSH_MSG_GLOBAL_REQUEST_SUCCESS
     string "tcpip-forward"
     uint32 port that was bound on server

  We should also explicitly allow binding to port
  0 to indicate any port.

Joseph Galbraith
galb-list%vandyke.com@localhost