- From: Jason White <jason@jasonjgw.net>
- Date: Tue, 13 May 2008 11:13:27 +1000
- To: list@html4all.org, "public-html@w3.org" <public-html@w3.org>, W3C WAI-XTECH <wai-xtech@w3.org>, wai-liaison@w3.org
On Fri, May 09, 2008 at 04:06:32AM -0500, Laura Carlson wrote: > We would love to have your input. Please send your comments to this > thread by May 22. A copy of the draft is also in the Wiki [3]. This proposal is much more consistent with W3C accessibility guidelines, developed over many years, than the text currently in the HTML 5 draft. As a result, it is more in line with best practice and policy in organizations that have adopted the Web Content Accessibility Guidelines, and which, in the time-frame for the development of HTML 5, will move toward the implementation of WCAG 2.0, which is now a W3C Candidate Recommendation. It should also be noted that these guidelines have been incorporated into policy, directly or indirectly, in a number of jurisdictions, and that future versions of HTML should be consistent with W3C accessibility guidelines as well as the practices that have emerged in support of them. My reservations regarding this proposal are as follows. It provides much non-normative guidance in the application of the ALT attribute, which may not be appropriate for inclusion in a markup language specification, and which moreover could be seen as usurping the role of WCAG 2.0 and its techniques documents. A format specification is not a tutorial. Nevertheless, there is a legitimate role for non-normative explanations in clarifying the normative content. I think the discussion in this case should be confined to a concise description, consistent with WCAG 2.0, of the function of @alt, a brief discussion of the various possibilities as outlined in WCAG 2, guideline 1.1, and a reference to that specification and its techniques for further details. Much attention has been paid to the syntactic question of whether @alt should be a required attribute. Ultimately, this depends on the question of what validating implementations of HTML 5 should treat as correct - what kinds of errors should be flagged by validators, whether operating as stand-alone applications or in authoring tools. Given the HTML 5 approach to error handling in user agents, it seems that regardless of how the syntactic issue surrounding @alt is decided, the specification will need to define graceful error handling behaviour, to be consistently implemented by user agents, in the case of a missing @alt. The present proposal does not address this issue. I think it would be helpful if the working group were to clarify, as a preliminary to addressing the syntactic issue of whether @alt should be mandatory, the role of validating implementations and the extent to which validity requirements should be designed to encourage, and to identify possible departures from, good authoring practices, including practices required by other W3C specifications such as WCAG 2.0. I would also suggest that consideration be given to the possibility of multiple levels of validation, of distinguishing, for example, fatal errors from warnings, as compilers do in parsing source code. It should also be borne in mind that accessibility-related validators play an important role in assuring conformance to those aspects of WCAG 2.0 that are testable at a syntactic level, and in notifying authors of potentially non-conforming content. Perhaps there needs to be a broader specification covering HTML validation that would encompass internationalization, accessibility and other authorial best practices, identifying aspects which can be tested by syntax alone and the kinds of warnings or errors that should be given in each case. These requirements would also apply to code generators and other authoring tools which impose syntactic constraints beyond those required for HTML parsing, on the document instances that they produce.
Received on Tuesday, 13 May 2008 01:19:29 UTC