[html4all] Figure ideas (was: Article by Catherine: Feedback on accessibility concerns in HTML5)

Thu Sep 20 16:43:16 PDT 2007

On 2007-09-20 16:03:00 +0200 zara <ecrire at catherine-roy.net> wrote: 
>  Joshua wrote: [...]
>> Interesting proposal. Is this along the lines of some possible uses
>> of the <figure> or <object> elements? A closer association of the
>> image and its alternate or long description would be a good thing IMO

> [...] The proposed "figure" element[1] was offered as a possible
> solution because it could be more explicit as to its function compared
> to "object" [...]

> [...] I wonder however if AT would rapidly support something like
> this. From what I was told during the initial discussion, it took a
> while for JAWS, for example, to support longdesc.

THe main challenge when it comes to compatibility wiht JAWS as well as graphic UAs, seems to be the text caption element. Here are some thoughts in that regard - which I hope to get some feedback on:

	QUESTION:  what is it that makes the association between the embedded element of FIGURE and the supplied text caption? Is it the very FIGURE element? Or is it something with the text caption element? I think it is the nature of the FIGURE element.

	PROPOSAL: To abadon, for semantic and compatibility reasons, the proposed LEGEND - and/or LABEL and CAPTION as caption text element in FIGURE. And instead: The text caption element should be one of the following phrase elements: SPAN, ABBR, CODE, VAR, SAMP, I, DFN, A  - depending on what is fitting in each context. (Or simply say that it should be SPAN - with the others as possible content of SPAN.) Those element is supported well by all UAs, and could also add semantics and useful features to the FIGURE element.

	SEMANTIC RATIONALE: The draft says that FIGURE is a paragraph element - and thus, it is not meant to solve the general problem of explicitly associating an embedding element and a text caption element - in any other way than the natural/implicit «paragraph association»: It is a specialised paragraph were only embedded content and an associated text is permitted. Thus I would maintain that the association is quite similar to the association between a DFN element and an paragraph - about which the HTML5 draft says:

        «The paragraph, description list group, or section that
        contains the dfn element contains the definition for the term
        given by the contents of the dfn element.»

Likewise, one could say that the text caption element of a FIGURE element defines the embedded element because it is the only allowed textContent element in a FIGURE paragraph. As such, it should be simple for e.g. JAWS to associate embedding element and text - as long as it understands that FIGURE is a paragraph.

I also don't think it is very fitting nor very logical to use LEGEND or LABEL from a semantic point of view - since those elements have quite spesific uses, quite distant from anything related to embedded content. 

	COMPATIBILITY RATIONALE: Those Catherine mentions with regard to JAWS. Those graphic UA issues which [Maciej] mentioned in his «Proposal for <figure> graceful degradation». The CAPTION element could have been a very logical element - it is also a an inline content element - and thus would fit well inside a paragraph element. However, the compatibility issuse speaks against it (not possible to style in current IE versions).

	FEATURE BENEFITS: HTML 5 draft says that «The dfn element enables automatic cross-references». Simply put I gather that it works this way: When the title attribute of the SPAN, ABBR, CODE, VAR, SAMP or I element equals the title attribute of a DFN element, then an automatic cross reference is created from those elements to the DFN element. Thus, by saying that the caption text should be enclosed in one of those elements - one could also simply create cross refrences to - let's say - long descriptions and etc. Thus, letting the caption text be a phrase element, would build on that cross-reference feature and probably help build support for it.

	I also think that not using LABEL, LEGEND, CAPTION would simplyfy things for authors - it would make it simpler to get that FIGURE is just a specialised paragraph element. And it would also bring more focus on the purpose of the caption text.

Please let me know what you think about these ideas.

[Maciej]: <http://www.w3.org/mid/[email protected]>
-- 
leif halvard silli