RE: UA <=> SS Protocol

(I'm struggling to come up with a  funny quip about Dave wearing a chair on his head, but haven't succeeded yet... :P )

Dave has a good point.  I'm not sure it's a slam-dunk, but we should discuss it.

I pretty much agree with Milan.  Browser-independence is critical and protocol requirements of the type we're discussing here are fundamental to that.  Furthermore, I *fundamentally* don't believe very many (any?) interesting and practical applications can be built without this, for all the reasons previously discussed over the last month, in a number of different requirements.

In other words, I don't think this XG has much practical value without being backed by a reasonable protocol.  I also believe we're the best group to identify the requirements for that protocol.  Whether we include the protocol specification in our charter, or have the same group of people work with the IETF (and if so, when), is worth discussing.

Personally, I think it's more pragmatic that this group make a recommendation based on our determination of requirements from the API discussion, and then determine whether it needs to be taken further with another group.

From: public-xg-htmlspeech-request@w3.org [mailto:public-xg-htmlspeech-request@w3.org] On Behalf Of Young, Milan
Sent: Friday, December 03, 2010 10:25 AM
To: Dave Burke
Cc: public-xg-htmlspeech@w3.org
Subject: RE: UA <=> SS Protocol

Hello Dave,

I'm interested to hear from the other chairs and Robert from Microsoft on this subject.  But personally I could be convinced to proceed in the direction you are suggesting.

I would like agreement, however, that there will be a tight coupling between the UA API specification and the protocol.  Referencing the WebSocket model, for example, the abstract at the top of the document states: "This specification defines an API that enables Web pages to use the WebSocket protocol for two-way communication with a remote host."

In the paragraph below we see similarly binding language: "This specification is being developed in conjunction with an Internet Draft for a wire protocol, the WebSocket Protocol..."

I believe such a yin/yang relationship is necessary to achieve the "browser-independent" experience suggested by our mission statement.  Concretely, when a specification is finally produced, no UA could call itself conformant unless it implemented the required portions of both specifications.


Thanks


________________________________
From: Dave Burke [mailto:daveburke@google.com]
Sent: Friday, December 03, 2010 5:19 AM
To: Young, Milan
Cc: public-xg-htmlspeech@w3.org
Subject: Re: UA <=> SS Protocol

<chair hat="on">

Within our XG, I think it's OK to suggest the existence of a protocol and possibly even specify requirements for that protocol. But defining/designing it neither fits into the purview of our charter nor in fact that of the W3C. Protocols for the Web instead belong in the IETF, and a possible route here is that one or more interested parties create an individual-contributed Internet-Draft in that space.

This is a well trodden path. For example, the WebSocket API is being standardized by the W3C and the WebSocket protocol is being standardized by the IETF. Similarly, for VoiceXML, a couple of like-minded W3C folks got together and wrote RFC 5552 to define the SIP protocol/interface to VoiceXML servers.

</chair>
On Thu, Dec 2, 2010 at 10:48 AM, Young, Milan <Milan.Young@nuance.com<mailto:Milan.Young@nuance.com>> wrote:
On the call today we agreed to split the protocol discussion from the web API.  This will both make it easier to read the document, and allow folks to easily filter out discussions which are not relevant to them.  But both sets of requirements will continue to reside in the XG document which Michael maintains.

Before getting into the weeds of the protocol discussion, I'd like to suggest deleting FPR30 which currently reads: "The communication between the user agent and the speech server must require a mandatory-to-support lowest common denominator such as HTTP 1.1, TBD."

I propose to replace this with a new requirement which will mark the start of the new protocol section.  The heading will read:  "User agents and speech services are required to support at least one common protocol."  And the description that follows: "A common protocol will be defined as part of the final recommendation.  It will be built upon some TBD existing application layer protocol such as HTTP."

Acceptable?