[rfc-i] FW: Accessibility and figures
masinter at adobe.com
Fri Apr 27 20:33:57 PDT 2012
> If in the future we have HTML with pointers to graphics, you would want that paragraph to be the alt-text? That seems unwieldy.
I don't think it needs to be alt-text. Putting it in the normal text content is not only fine, it's usually better. That way, the document makes sense without the graphics, or without actually being able to look at the graphics.
With regard to Figure 1 of http://tools.ietf.org/id/draft-iab-rfc-editor-model-v2-05.txt
>> Now, the diagram has marginally more information:
> No, it has much more information than the paragraph. The amount of difference is relevant for this discussion.
>> You don't have to read RFC 4844 to see what the four existing streams of documents are.
>> There are links in the diagram that aren't described in the text, but I don't think those links
>> are very useful (it's suspicious to have a box labeled "Community at Large" as if it were a single
>> entity), and many of the diagrammed relationships are not apparent from the diagram (ISOC and RSOC).
>> But the diagram is there for illustrative purposes, and all of the relationships that are interesting seem
> > to me to be described in the text or at least in referenced documents.
> ...which is exactly the point of this thread. Are you expecting the accessibility text to be a full description of the contents of the figure? If not, is a simple summary good enough?
Not a 'full' description, but 'sufficient information in the text that the figure is not necessary to comprehend the essential points of the document. But 'a simple summary' is not necessarily 'good enough', unless the entire illustration is just of an example which isn't essential, just helpful (a reader would have to spend more time puzzling out the text and drawing their own diagrams, for example.)
> Most people on the list appear to agree that figures are normative, regardless of whether they are text-art or graphics-art.
> Is the accessibility text you say we need also normative?
> If not, a sight-impaired reader will not be able to rely on it. If you are adding accessibility text to make the spec implementable
> by a sight-impaired reader, the accessibility text needs to describe the contents of the art completely and accurately.
Note that in this case and in many others, the text isn't alt-text (it should not be a "non-sighted alternative" which isn't visible to sighted users) but rather part of making sure the diagram isn't the ONLY representation of important information and constraints".
> I propose that accessibility text should be allowed to be descriptive and not complete, and should be considered non-normative.
I'm not going to get too excited by "normative" vs "non-normative".
For formal specifications, given the review process and the assumption that someone reviewing the document had ready access to all of the content being reviewed, alt-text is at greatest risk.
I think the reality of looking at a spec is deciding whether the (other) implementors of the spec agreed with what the spec says. Alt-text is less likely to be reviewed by everyone, so should be treated with suspicion. But if alt-text doesn't match the image, it's a spec bug, and it's still open as to which one is wrong. Setting a precedence doesn't help.
More information about the rfc-interest