Re: [Stox] Communication between STOX-capable and non-STOX-capable entities

[Matt Ryan <[e-mail]>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Further comments in line.


On 5/29/14, 8:48 AM, Matt Ryan wrote:
> Saúl and Peter, thanks for your input on my question.
> 
> I agree that this is the approach we should take.  This direction
> is supported by RFC 7427 (previously draft-ietf-stox-core) where
> the responsibility for determining the format for communication
> and translating if required is put upon the sending server and
> it's accompanying gateway, and not upon the receiving server.
> 
> Because of this, I have further concerns with
> draft-ietf-stox-chat-06 as well as with
> draft-ietf-stox-groupchat-04.  I will enumerate these each
> separately in their own review emails.

For simplicity's sake let me just enumerate my general suggestion
instead of separate review emails for each pertinent draft.  My
apologies for being indecisive.


I suggest the drafts contain some formal language indicating the
assumptions that should be followed on each side in order to emphasize
the interoperability point.

For example:
- - Given a server sending a message to another server with a different
protocol, the sending server doesn't know whether the receiving server
supports STOX, and interoperability is a goal, so there should be
language indicating that the server SHOULD (MUST?) support sender-side
translation of messages from one protocol to the other.
- - Because a receiving server might be receiving messages from a sender
that does not support STOX, the receiving server SHOULD support
destination-side translation of messages from one protocol to the other.

I believe this is in harmony both with RFC 7247 and the clarification
I received in answer to my question, and that such language would both
support and emphasize these two important points.


> But in summary, the concerns are that the examples given in these
> drafts show the gateways on both sides of communication as being
> involved in exchanges where they should not be required, and that
> showing the involvement of both gateways gives the impression that
> both ARE required.
> 
> In other words, it does not emphasize the point that STOX-capable 
> servers must be able to communicate effectively with
> non-STOX-capable servers, which it seems we agree is important.

I believe the discussion in a separate thread addresses this concern
for me, wherein we've talked about reworking the examples in the
drafts to not show the receiving-side gateway in the diagrams.



> 
> 
> On 5/29/14, 6:46 AM, Peter Saint-Andre wrote:
>> On 5/28/14, 6:37 AM, Saúl Ibarra Corretgé wrote:
>>> Hi Matt,
>>> 
>>> On May 20, 2014, at 4:36 AM, Matt Ryan wrote:
>>> 
>>>> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512
>>>> 
>>>> Should a STOX-capable server be able to communicate with a 
>>>> non-STOX-capable server, so long as the communication is 
>>>> started by the STOX-capable server?  Is this among the
>>>> current goals of the WG?
>>>> 
>>>> Sorry if this has been covered - but this particular point 
>>>> remains unclear to me despite numerous readings through 
>>>> draft-ietf-stox-core and some of the other supporting
>>>> drafts.
>>>> 
>>>> 
>>>> Let me provide a simple example to explain what I mean.
>>>> 
>>>> Suppose an XMPP client initiates a chat to a user that
>>>> happens to be a SIP client user.  Per draft-ietf-stox-chat,
>>>> this request flows through the XMPP Server which determines
>>>> that the message is targeted to a SIP user, and then invokes
>>>> its own XMPP-to-SIP Gateway to translate the request to a SIP
>>>> request and send that request on to the appropriate SIP
>>>> server.
>>>> 
>>>> So far, the XMPP Server knows that the target user is a SIP 
>>>> user, but does not know anything about the SIP Server 
>>>> capabilities other than that it is (presumably) a SIP server.
>>>>  Importantly, the XMPP Server does not know whether the SIP 
>>>> Server also supports a SIP-to-XMPP gateway.
>>>> 
>>>> 
>>>> This is where my question comes.  Given the described XMPP 
>>>> Server with a corresponding XMPP-to-SIP gateway, is it meant
>>>> to be the case that this XMPP Server should be able to
>>>> communicate with any valid SIP server, so long as the
>>>> conversation is initiated by the XMPP Server?
>>>> 
>>>> 
>>>> This is an important point because it has direct impact on 
>>>> several of our current drafts.
>>>> 
>>> 
>>> The idea is that a STOX capable infrastructure is transparent
>>> to others. To the eyes of the world,  given domain, say
>>> example.com has SRV records for SIP and XMPP. The fact that
>>> there is an actual gateway handling the traffic and doing some
>>> translations is a detail of that infrastructure. So a non-STOX
>>> server should be able to communicate with a STOX server just
>>> fine.
> 
>> +1
> 
>> Peter
> 
> 
> 
> 

- -- 
Matt Ryan
Code Slinger  |  Jive Communications, Inc.
Jive.com  |  mryan at getjive dot com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.22 (Darwin)
Comment: GPGTools - https://gpgtools.org
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBCgAGBQJTjK7sAAoJEPyFiw1u4O0cT3oH+gPsBVdEARpG8S81LMInTAzT
vPmJMySq0T9fdCm5F0CLudEp9+tfZgGDg1DRCtBYRZWn8AhxK3GCUdlkT0OjO1a7
v4Ordp8dtJNcVLsHsQw7WJX5gkfJGgNrmJ2q1mIsNy1m2YtmGRvTQ6qY1duMUvQu
UuIJEIk5eIvOYIy7VhGUsnOMJmvXnxLSO5Lg6KKQNKtumU4VLduYdDTPyr4t6pk8
MIwmbJsbxlD3x3ltBvrnCVey6NUhWac1YcKFOediYeb/7LlaPfkkA4zoD2k0pyfH
4HFuYoqnNyDp3gohzSgDo+zAITgInv4YMeP5bmiNhaeIKGFZ5jAmUsfUUus7f2Y=
=WNXx
-----END PGP SIGNATURE-----