Re: [Stox] Pete Resnick's IESG feedback on draft-ietf-stox-core

[Peter Saint-Andre <[e-mail]>]

On 2/10/14, 5:11 PM, Peter Saint-Andre wrote:
> On 2/10/14, 3:08 PM, Peter Saint-Andre wrote:
>> Pete, you wrote:
>>
>>    Section 4 is (nicely) clear on the document architecturally
>>    describing a gateway. However, traditionally a gateway is
>>    transparent to the entity that communicates with it: When we
>>    had SMTP-to-X.400 gateways, the gateway appeared as just another
>>    SMTP system that noticed special qualities of the address and
>>    then routed the messages appropriately.
>>
>> Your "traditionally" sounds like an SMTP phenomenon. In XMPP, we don't
>> have intermediate transfer agents, and XMPP servers are designed to use
>> add-on modules for additional functionality. So in the XMPP universe it
>> makes sense for the operator of an XMPP service to install a module (in
>> addition to the XMPP server) that performs the gateway function (we
>> often call this a connection manager).
>>
>>    Section 5 describes
>>    something a bit different. It's not clear that what section 5
>>    describes actually is part of the gateway, but rather sounds
>>    like a combined server which does failover between the
>>    protocols. I don't think this is a showstopper, but it might
>>    help implementers significantly if you described in section 5
>>    *where* in the model this function occurs. Right now, it
>>    sounds like the server itself does the addressing failover,
>>    not the gateway.
>>
>> Yes, I see your point. This kind of thing is quite likely implementation
>> specific (e.g., when the add-on XMPP-to-SIP gateway module gets
>> configured into and trusted by the core XMPP server, it might get added
>> into an event listener for core stanza delivery if an XMPP lookup fails
>> for the remote domain). Let me look at what text might be useful here.
>
> How is this for proposed text?
>
>     Existing SIP and XMPP server implementations do not typically include
>     the ability to communicate using the other technology (XMPP for SIP
>     implementations, SIP for XMPP implementations).  One common
>     architectural pattern is to associate a gateway with the core server
>     implementation (e.g., in XMPP such a gateway might be called a
>     "connection manager").  How exactly such a gateway interacts with the
>     core server to complete tasks such as address lookups and
>     communication with systems that use the other technology is a matter
>     of implementation (e.g., the gateway might be an add-on module that
>     is trusted by the core server to act as a fallback delivery mechanism
>     if the remote domain does not support the server's native
>     communication technology).

Pete, that text is now in draft-ietf-stox-core-10.

Peter

-- 
Peter Saint-Andre
https://stpeter.im/