Re: [xmpp] Fwd: Reviews of draft-ietf-xmpp-websocket-07

[Dave Cridland <[e-mail]>]

On 4 July 2014 19:02, Matt Miller <[e-mail]> wrote:

> On 7/3/14, 9:59 PM, Lance Stout wrote:
> > 1. A server SHOULD use <close see-other-uri="..." /> when ending a
> > session & pointing the client to a new endpoint. Was previously a
> > MAY, but IETF LC feedback pointed out that we don't have any other
> > defined way to do this.
> >
> > I don't feel too strongly on this point, so feedback welcome on if
> > this change is really needed.
> >
>
> I agree with this change.  Making the 'see-other-uri' behavior a
> SHOULD I think will lead to more interoperable software; people will
> be much more inclined to implement it.
>
>
I'm not entirely sure that 2119 language makes sense here for the server.

If a server wishes to tell a client to switch websocket endpoint, then this
is the way to do it. If it doesn't do it this way, then it can't tell the
client. There's no "SHOULD" or "MAY" or "MUST" about this, because there's
no alternative.

The interesting question is what clients are expected to do here - clients
MUST reconnnect to the other URI? clients MAY? clients SHOULD? There's no
description of the clients' behaviour here at all.

The other question that lurks around my head is the interaction with
XEP-0198 resumption here.

Dave.