Re: [html-techs-tf] caption vs alt

Hello all,

Sailesh wrote:

 > As I see it, the figure element is a tool for supporting SC 1.3.1 and
 > the figcaption's  primary purpose is to label the figure;  that it can
 > be used to label an image within is secondary if not incidental.

I definitely agree, the <figcaption> identifies the <figure>, not the 
image or any other content of the figure (for example, if the figure is 
a code example). Maybe the image can take the accName from the 
<figcaption> as a "fallback" in case there is no @alt, but this is also 
an assumption that the <figcaption> serves the same purpose as the image 
itself, which in my opinion would not be so common. In any case, the 
image must be exposed to the SR, so suppressing the @alt makes no 
difference in terms of what is read.

The only advantage I can see would be the few bytes of the @alt in the 
HTML code, and maybe better consistency if both the @alt and the 
<figcaption> are created manually ans separately. I guess that in most 
cases they will be generated through a CMS.


 > If there were a batch of pictures uploaded with no ability to add alt,
 > I suppose they should all be in a single figure element (as a group)
 > or they should be marked up with li in a single ul/ol.

Indeed, I see a risk here. Imagine that we include a single figure for a 
group of images that belong to the same event:

<figure>
   <img src="csun14_01.jpg" />
   <img src="csun14_02.jpg" />
   <img src="csun14_03.jpg" />
   <figcaption>CSUN 2014</figcaption>
</figure>

Would "CSUN 2014" be a valid text equivalente for every image? How can a 
screen reader user distinguish between them? Shouldn't be this case 
explicitly prohibited? Would we allow 2 of the images to have an alt ant 
the third one to rely on the <figcaption> (assumming that it would be a 
valid text equivalent)?

> I do not see why it is critical to:
> - group an image with a grouping element when the figure has no other
> content than an img and a figcaption. What is being grouped together?

Maybe it is not critical, but there are some use cases. For example, you 
can write several articles with illustrations and use CSS to 
automatically number them in the <figcaption>:

main { counter-reset: figs; }
figure { counter-increment: figs; }
figcaption::before { content: "Figure " counter(figs) ". "; }

Then, you can reorder or reuse the illustrations preserving an 
appropriate order.

> - and to grant an exception for not using alt only in this situation.
> It simply complicates matters for developers and testers.

I agree.


> Ramón: In my experience JAWS14/15 (with FF) announces both figure and
> the div with role=group  as a group and not as a frame.

I've just re-tested and you are right... In English. I originally tested 
in Spanish and JAWS reads it as "marco" (= frame), so I assummed that 
the translation would be that. This is also the case for any 
role="group" element, so I will file a bug to the Spanish translation of 
JAWS.

Regards,
Ramón.

Received on Wednesday, 15 January 2014 13:11:28 UTC