Re: Terrapin

[Mouse <mouse%Rodents-Montreal.ORG@localhost>]

>> If the sequence number is not usable for its intended use, then,
>> unless someone has a use it _is_ usable for, it should be fixed or
>> eliminated.
> The sequence number is also used in MSG_UNIMPLEMENTED to refer to the
> unimplemented packet.

Fair point.

The question then becomes a tradeoff: is this use of the sequence
number as currently defined useful enough to make it worth keeping?
I'm not sure what I think of that.  But then, I don't like
MSG_UNIMPLEMENTED to begin with; an unrecognized message type byte, to
me, indicates a severe bug in at least one end.  I consider tearing
down the connection a much more appropriate response in general.
That's what my implementation does.  (If-and-when peers start throwing
such atrocities at me, I'll have to figure out whether to add a
misfeature-compatibility workaround or write off interoperability with
them.)

And I _really_ don't like the use of unrecognized packet type bytes,
depending on them to produce UNIMPLEMENTED responses, as a Terrapin
workaround.  Besides a general dislike for abusing a misfeature, the
result is incompatible with anyone who's using currently-unused packet
type fields for any other purpose.  Sooner or later something will get
defined to use one of them, at which point this Terrapin workaround
will explode badly - or, worse, the presence of the workaround in the
SSH ecosystem will make it effectively impossible to introduce any such
new feature, forever crippling SSH extensibility by effectively
rendering all currently-undefined type byte values permanently
undefined.

> [T]heir approach imply adding an other magic symbol in the kex
> algorithms list, that's why I prefer to add new algorithms.  The
> intent of the specification is not to have some magic algorithm names
> that must never appears in the other direction.

I agree.

> I know that we already did it with extinfo for good reasons, to work
> around broken implementations that would otherwise disconnect, if I
> remember well.

I don't consider that a good reason.  I firmly believe any
implementation so broken as to disconnect upon its peer trying to use
the defined extension mechanisms should be rendered _obviously_ broken,
so as to get it fixed or replaced.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse%rodents-montreal.org@localhost
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B