Re: unifying alternate content across embedded content element types

At 22:28 -0500 UTC, on 2007-07-14, Jon Barnett wrote:

> On 7/14/07, Sander Tekelenburg <st@isoc.nl> wrote:

[...]

> I'll make a habit of sending text/plain [...]

Thanks. Much appreciated.

>> > [...] context is often equal to fallback. There are cases where the
>>context
>> >and the fallback would be exactly the same
>>
>> Could you give an example? I'm having a hard time thinking of one...
>
> <img src=cat.jpg alt="A photo of my cat, Fluffy, playing with a ball of
>yarn">
> <p>A photo of my cat, Fluffy, playing with a ball of yarn</p>

Here the paragraph functions as a caption. I'm not sure that counts as
"context" actually... Is the idea that this is the entire content of a single
HTML document? If not, what is the context? Is it a page about Fluffy, about
yarn, about photography, about you, about play, about HTML, about english,
about cats, about what?

That aside, the alt text is most likely[*] just plain wrong. See further
down. (And see several previous messages in various threads these past
months, where different people attempted to explain what "correct" alt text
is.)

[*] I say "most likely" because these things are impossible to judge without
a context. The exact same picture can have a completely different meaning in
different contexts. Off the top of my head, the only context I can think of
in which it would be correct would be a page about HTML, explaining proper
alt text, using this as an example of improper alt text.

> figure/legend helps with this by directly associating the photo with
> the description Expecting me to provide that @alt attribute along with
> the following paragraph seems silly and redundant.

Indeed. And that's not what any version of the HTML spec expects you to do.

But it's quite clear that what authors *are* expected to provide through @alt
is not clear to most of them, so there would seem to be room for improvement
of the spec to make that more clear. (I don't think there's any real
disagreement on that, except for whether this should be in the spec itself,
or in a separate tutorial.)

> If that's not what
> I should provide for @alt, then it should be crystal clear what I
> should provide as @alt

As an author, you should look at the entire document, without the images
being loaded, and for each image consider what text would make sense in its
place; what text would make you not miss the image, because that text conveys
the same as the image.

If, using that approach, your alt text doesn't look right to you, it isn't
right. If you can't come up with alt text that truly tells what you intend to
tell through the image, then alt="" is much better than alt="something
useless".

Personally I think this method is quite clear. Just as it is clear that it
will often not be easy at all for authors to come up with the right alt text.
I often find it quite hard.

[...]

>> Sorry, you've lost me. I don't know what "long vs short" refers to, and
>>don't
>> understand what "semantic equivalent vs semantic alternate" means.
>
> Most of this thread has been about various ways of providing "short"
> alternative in addition to a "long" alternative, such as using @alt
> and @longdesc together, and using a hypothetical <object alt='...'>
> attribute in addition to the contents of alt.  I thing this confuses
> the issue of alternative content more than it goes to solve it.

Those things have come up in this thread, but the subject is how to unify the
mechanisms for authors to provide equivalents. In other words: can we come up
with one single mechanism, for every relevant element, that allows authors to
provide equivalent content? (*And* have it be backwards compatible with
pre-HTML5 UAs.)

(But yes, there's certainly some confusion in this thread . We all try, but
seem to have a hard time understanding each other now and then.)

> I think the more relevant issue is "descriptions" of media vs.
> "equivalents" to media.

You've lost me again ;)

> Were the examples I gave not clear enough as to what I meant?

Speaking for myself: I couldn't follow what you meant to say with them, no.

I'm still not entirely sure of what you mean, to be honest ;) Are you
basically asking what proper alt text is? Or are you posing that the HTML5
spec needs to make it easier for authors to understand what proper alt text
is?

[...]

> I suspect that most commonly in the wild, the contents of <object> are
> actually a description of the media instead of an actual equivalent to
> the media.

Might well be that that's as abused as @alt, yes. Wouldn't surprise me.

> It would be interesting to gather data on this.

Indeed.

> Is there
> a convenient way for me to search the code of existing pages on the
> web.

Define "convenient" :) To judge whether the alternate content is truly
equivalent content you'd need to look at both, and their context. I don't see
how you could do that programatically.


-- 
Sander Tekelenburg
The Web Repair Initiative: <http://webrepair.org/>

Received on Sunday, 15 July 2007 05:16:21 UTC