RE: closing a channel

["denis bider" <ietf-ssh%denisbider.com@localhost>]

> > What do you think about this proposal:
> > link CHANNEL_EOF to all of CHANNEL_DATA,
> > CHANNEL_EXTENDED_DATA, and CHANNEL_REQUEST,
> > and create two new channel message types:
> > CHANNEL_DATA_END, linked only to CHANNEL_DATA,
>
> I don't like it. It's too much extra complexity
> for a small very gain. It's the kind of improvement
> that identifies a design by a committee...

Strongly agreed.

At this point of the protocol's evolution, drastic solutions like
introducing new message types would be counter-productive. All that is
required is a clear definition of what exactly CHANNEL_EOF means. This
definition should be compatible with most existing SSH2 implementations.
(Except obviously broken ones.)


> Then enters the EXTENDED_DATA hack,

I'm not bothered by EXTENDED_DATA being a hack. It is what it is, it works
how it works, that's fine with me. Now that the cat is out of the bag, and
50% of the world's secure shell servers already support SSH2, we won't get
rid of EXTENDED_DATA by redesigning the protocol. If we try to do so, we'll
just have to support the new system as well as the existing one.