- From: Scott Luebking <phoenixl@netcom.com>
- Date: Mon, 22 Nov 1999 18:39:23 -0800 (PST)
- To: charles@w3.org, phoenixl@netcom.com
- Cc: w3c-wai-gl@w3.org
Hi, Charles I believe your analysis is not very complete. For example, results of search engines cannot be generated before the search query is specified. Or a web page which has the most recent article from some set of web sites listed in a user specified profile. Or web pages which are stored at a central repository and generated on demand depending on the university which is the subscriber. Or a web page listing descriptions of items in a shopping cart. Or a web page which does a comparison of long distances costs depending on a user's calling pattern. In terms of templates, you're making certain assumptions which aren't necessarily true. Templates can have multi-layers of flexibility when structured proerly. You're assuming that the templates must be discreet from each other. In a well architected system, it is very easy to specify structures which can be shared among templates. This allows for a single change to be easily propogated. This approach can actually allow for multiple formats outside of 2. The trick is to think abstractly about the attributes. (If you're a programmer, I can show you software which does that.) Scott > Actually there is no need for the document to be generated at request-time: > there are significant benefits in speed from generating a ste from a database > and making it a static collection, even with multiple formats. > > The generation of only 2 forms does not avoid the problem of compromising the > accessibility of a page for a person with low vision who requires on both a > structured page and visual cues to the structure (and does not address the > needs of people who have mobility impairments, etc.) In fact to support this > approach multiple combinations are required. The amount of work generating > mulitple combinations from a database is in effect the amount of work to > multply the templates, and the amount of work maintaining consistency across > multiple templates when there is any change required is substantial, and > anecdotal evidence suggests it is not done well. > > Charles
Received on Monday, 22 November 1999 21:39:25 UTC