- From: Gregory J. Rosmaita <oedipus@hicom.net>
- Date: Thu, 29 Mar 2007 15:26:52 -0500
- To: Maciej Stachowiak <mjs@apple.com>, public-html@w3.org
Maciej Stachowiak wrote, quote: Thus, visible metadata reduces the need for browser involvement in web innovation. unquote whilst i understand your argument maciej, visible is in the eye, ear, fingertip or whatever else is available to the user. the point of the discussion, as i read it, between invisible and visible metadata would be better threaded as: perceptible versus hidden metadata -- that is, metadata which isn't necessarily exposed, but which provides a programmatic option for: A. user cognizance of the presence of the hidden metadata, without which exposition of the hidden metadata, is impossible; B. knowledge of the existence of, and access to, the otherwise imperceptible metadata; no matter what modality through which the document has been parsed and presented to the user, the intended functionality of hidden metadata depends upon a programmatic cue from the user agent, which can either be communicated directly to the user interacting with the user agent or processed as a programmatic switch which then effects, and is affected by, user defined settings and options, as well as triggering appropriate switches in third-party software, such as a screen-reader. which is why the A element has become, by default, and may well remain, the catch-all element for exposition of otherwise hidden metadata, targets, or options: it remains the only ubiquitously supported means of pointing to a target, such as a longdesc page, and enabling the user to obtain the target document or other associated metadata resources. i had to use the A element to encase an IMG declaration which contained an identical target, pointed to through use of the LONGDESC attribute, before mainstream assisstive technologies began to recognize and support IMG elements for which LONGDESC has been described, but i shouldn't have had to wait so long for an assisstive technology vendor to implement support for an interactive attribute such as LONGDESC -- support for LONGDESC (which can provide meta- and micro- data) should be available to any users of all user agents, as part of their conformance to the specification... yes, i associate the LONGDESC explicitly with the IMG it describes, but i also use A extensively (admittedly as a hack) to make the IMG element an actionable item, even though that is what LONGDESC is intended to provide the user in a transparent manner - that is, recognition of, and support for, invisible metadata.which can be made quote visible unquote - or, rather, perceptible - to the tools which the user is currently using; optimallly, LONGDESC would be universally supported at the user agent level, for not everyone who needs or wants a LONGDESC will be using a screen reader or other assisstive technology, so the option to access quote invisible unquote metadata or additional information about an IMG, OBJECT, or other actionable elements (such as Q) must be presented to all users at the user agent level, upon user request, but the request cannot be made if the user has no means of first being informed of the existence of the metadata defined for the document, as well as a means of accessing the specified target resource. in the end, the backplane mechanics should provide the user with a choice: automatically detect changes in lang and take appropriate, user-defined action -- such as either automatically switching to a more suitable natural language (for the purpose of visual, aural, and tactile rendering, based on user defined client side preferences, which i believe to be raman's prefered method), or setting the user agent to generate a choice of perceptible and actionable options. (which could take the form of flags automatically associated with an actionable icon - and appropriate ALT text - to allow for manual choice, in this case, of a more suitable natural language.iteration for a particular user. gregory j. rosmaita eddress: oedipus@hicom.net or gregory@ubats.org skype: oedipusnj sites: http://www.hicom.net/~oedipus/ http://ubats.org http://my.opera.com/oedipus/
Received on Thursday, 29 March 2007 20:27:08 UTC