W3C home > Mailing lists > Public > public-dcci-editors@w3.org > August 2007

Feedback on the LCWD of the DCCI specification from the Multimodal Interaction WG

From: Deborah Dahl <dahl@conversational-technologies.com>
Date: Thu, 16 Aug 2007 12:37:48 -0400
To: <public-dcci-editors@w3.org>
Cc: "W3C Multimodal group" <w3c-mmi-wg@w3.org>
Message-ID: <003801c7e023$c815af50$6701a8c0@chimaera>

The Multimodal Interaction Working Group has prepared the 
following feedback on the Delivery Context: Client Interfaces specification
Last Call Working Draft. Please feel free to contact us for clarification
on any of these points.

Thanks to Raj Tumuluri of Openstream for preparing this feedback.

best regards,

Debbie Dahl
for the W3C Multimodal Interaction Working Group


1. Registration for events:

1.1 Notification of Events ( Section 1.1 Uses for DCI ):

"...It also provides notifications when properties change.."

While the specification describes use of DOM3 events, given that DCCI will 
have a different root-element ( namespace), it is not clear how components 
that want the notifications register for the same..


MMI Architecture allows Interaction Manager (IM) and Modality Components( 
MC) to be distributed. And except in the case of nested-modality components,

modality components communicate with each other only through the Interaction

Manager. Given this principle of MMI architecture, MMI authors would be 
required under the proposed DCCI spec,  to implement another DCCI-interface 
component to register and obtain the local DCI events to pass on to IM for 
every device on which any modality component is running..

For example, a device-client could be getting the text-to-speech streamed 
from a TTS server. Now, if the user mutes the speaker on the device, an 
event gets generated on the local device through DCI, but the IM running on 
the server (a different device) will not get this event to signal the 
TTS-component to stop streaming audio, as it will not have a way to remotely

register and get this event from the device. So, under the proposed 
specification, an MMI author would be forced to implement another MMI 
component just for passing on DCI events to the IM. Please, note that the 
local Modality Component cannot directly do this job, as the local Modality 
Components are to be implemented as black-boxes and as such cannot snoop on 
these events and determine which ones should be passed-on to the Interaction


We would like to hear from DCI-WG on how this remote-registration for events

could be done, under the the current DCI framework.

2. Section 2.1 of DCCI specification on Interfaces :

"....DCCI is an interface that focuses on making properties from the 
delivery context
available to code executing within a web client [GLOSS]...

The glossary referenced above DOES NOT contain definition of web-client, but
client as described below :

Client ( www.w3.org/TR/di-gloss)
    The role adopted by an application when it is retrieving and/or 
rendering resources or resource manifestations.
    This term was taken verbatim from Web Characterization Terminology & 
Definitions Sheet.

Further, from MMI perspective clients needs not be web-clients ( meaning 
implementing HTTP protocol for communication with servers..)
Received on Thursday, 16 August 2007 16:38:24 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 19:35:16 UTC