W3C home > Mailing lists > Public > www-archive@w3.org > September 2007

Re: alt attribute comments?

From: Anne van Kesteren <annevk@opera.com>
Date: Sun, 23 Sep 2007 18:56:10 +0200
To: "Steven Faulkner" <faulkner.steve@gmail.com>
Cc: www-archive@w3.org
Message-ID: <op.ty39bwhp64w2qv@annevk-t60.oslo.opera.com>

On Sun, 23 Sep 2007 17:00:11 +0200, Steven Faulkner  
<faulkner.steve@gmail.com> wrote:
>> The HTML5 draft does not allow no alt attribute for any image. It's
>> actually a MUST level requirement to provide it for most and a SHOULD  
>> for images with critical content.
> from the draft spec the reason you give for not providing an alt text  
> does not appear to fit
> "the alt attribute should only be omitted when no alternative text is
> available and none can be made available, e.g. on automated image gallery
> sites."
> http://www.whatwg.org/specs/web-apps/current-work/multipage/section-embedded.html#the-img
> You could have provided alts for the images, but chose not to.

I suppose I could have, though my site does fit in that category. (It's  
automated and does not support insertion of replacement text at the  
moment.) It's also not really clear to me what description would be  
adequate enough, but that's a separate issue.

> This would take too much time for little benefit" then their pages will  
> be  be
> both valid and conformant.

Correct. Note that this would be true for <img alt=""> as well except that  
there the end user does not know there's an image at all on the site and  
therefore can't ask software or maybe a real person to describe the image  
for him.

> BTW Don't you think it a little odd that the draft
> tells authors that they MUST have the alt attribute  on meaningless  
> images but makes it optional on those with meaning that is critical  
> content?

This seems to fit with the way <img> has been designed and is implemented.  
I'm not really sure if anything else is possible unless there's an  
algorithm that can tell you whether an image is meaningless or not and I  
don't think that's possible.

>> Regarding no alt= attribute, that indeed seems to be case. Which is why  
>> we might want to try something else instead. On the other hand, going  
>> out
>> from the current state of the art of screen readers doesn't really help
>> solving this problem I think. Dunno really.
> yes i think something else is definitley needed. something to explicitly
> flag images that have the alt attribute omitted for those reasons that  
> are cited in the spec.

Hopefully image interpretation software gets better soon. That will  
hopefully solve the real problem here which is that a lot of authors don't  
really have an incentive other than "search engine optimization" to  
provide replacement text for images. (I'm not a big believer in requiring  
such things by law, etc.)

Anne van Kesteren
Received on Sunday, 23 September 2007 16:56:28 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:33:15 UTC