- From: Roger Gimson <roger.gimson@hp.com>
- Date: Thu, 18 Aug 2005 18:05:33 +0100
- To: www-di@w3.org
- CC: Rotan Hanrahan <Rotan.Hanrahan@MobileAware.com>, Johannes Koch <koch@w3development.de>
I am probably responsible for creating the confusion here, as I believe I drafted the original definition of 'delivery unit', as it appears in the September 2003 version of Device Independence Principles [DIP2003]: http://www.w3.org/TR/2003/NOTE-di-princ-20030901/ Example 2.2.1.2 in [DIP2003] suggests that a delivery unit is the response to a single HTTP request (whether direct or indirect in Rotan's description). The 'set of material' refers to the combination of representations that may be delivered together, such as HTML with embedded ECMAScript. However, Example 2.3.1.1 in [DIP2003] says that 'a user agent may render the material received in a delivery unit' - implying that it consists of all the material needed to render it, as Rotan suggests. This latter example, however, does make clear the distinction between a 'delivery unit' and a 'perceivable unit', which was the original motivation for introducing the two terms. Together, they replaced the term 'presentation data' used in the September 2001 DIP, which itself caused confusion: http://www.w3.org/TR/2001/WD-di-princ-20010918/ As far as device independence is concerned, the point is that adaptation may take place server-side, producing a delivery unit, and/or may take place client-side, producing a perceivable unit. Though it is not stated explicitly, the implication is that several delivery units may be required to produce one perceivable unit and that one delivery unit may produce several perceivable units. This is also the reason that the definition states that 'Users are not normally aware of individual delivery units.' Roger -- roger.gimson@hp.com | HP Labs, Bristol BS34 8QZ, UK | +44 117-312-8167
Received on Thursday, 18 August 2005 17:07:21 UTC