W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > January to March 2003

Re: ALT as required attribute

From: Bill Mason <w3c@accessibleinter.net>
Date: Tue, 04 Feb 2003 23:58:35 -0800
Message-Id: <>
To: w3c-wai-ig@w3.org

At 08:07 AM 2/4/2003, Jim Ley wrote:
>"Bill Mason" <w3c@accessibleinter.net> wrote in message
> >But it does clearly say that the attribute
> >is to "specify alternate text to serve as content when the element
> >cannot be rendered normally."  Thus in scenarios where the image
> >has rendered normally, the ALT content should not be presented
> >by a user agent in any fashion, tooltip or otherwise.
>That would suggest an ALT text of "This image will be available tomorrow" is

Unless that is the message that the image itself is meant to convey, then 
that would not in fact be OK.

>  (it doesn't mention users ability to percieve the image, purely the UA's
>ability to render it, I don't think anyone would agree with that.  UAAG
>clearly states that all content of all types should be made available to the
>user including showing alternatives at the same time.

I do not see this in the spec unconditionally.  Rendering the image and ALT 
content simultaneously is *one* option, not *the* option (guideline 2.3 
section 2). I do also see:

"Render content according to format specification" which includes "when to 
render the alt attribute in HTML".  Nothing you have said to date proves 
that a tooltip is a "when" to show ALT content, and the HTML specification 
is clear that a tooltip is not appropriate.

"Allow configuration to provide access to each piece of unrendered 
conditional content"
Unless your browser cannot turn off images, or view properties of an image 
through a right-click (or similar mechanism), you already have access to 
the ALT content.

>So there's plenty of support, and as a user who requires this, I fail to 
>see how you can argue
>against it?

Among other things, the specs you cite to support your position explicitly 
refute your stance.

> >If the user wants to find a way to make it so, whether via plug-in or
> >some advanced CSS in a user style sheet, good for the user.
>If the user is perfectly justified in doing it, then the user agent is also
>perfectly justified in providing it by default, the default
>rendering/configurations are just some arbitrary defaults designed to best
>serve the users, you seem to be saying that because the showing of ALT on
>tooltip does nothing to serve your needs the default is wrong, I can't see
>how you're justified in coming to that conclusion.

Because it violates the spec.  Your argument seems to itself come down to 
"the browser should do this regardless of specs because it does something 
to suit my needs".

> > At 02:14 AM 2/4/2003, Jim Ley wrote:
> > >Why are specifications mandating how semantic mark-up be rendered,
> > >that just make it visual mark-up?
> >
> > The specification is not mandating how.  It's mandating when, where ALT is
> > concerned.
>I thought you just agreed it mandated no such thing,

Obviously I did not agree to that, since your quote of me says the exact 

>and a UA (a browser+plugin or CSS) would still be compliant to the 

Of course the UA would still be compliant.  No UA could ever claim 
compliance if you're about to argue that if you can install a plugin that 
does something nonstandard, the UA is now out of compliance.  Do you 
suggest that the UA prohibit the user from installing a plugin if the 
plugin allows some noncompliant ability/activity?  UAs should police what 
gets installed on a computer system?

ALT is part of the HTML spec.  CSS is another specification entirely.  If 
there is no tooltip for ALT, that obeys the HTML spec.  If you write CSS 
rules to force-display the content of ALT with all images in your browser's 
user style sheet, that obeys the CSS spec.

You can't have it both ways.  Do you want a means to display ALT regardless 
of the image's status (that falls within existing specs)?  Or is your 
argument now that if a browser supports CSS2, it is in violation of the 
HTML spec?

> >>Why, I'm pretty rare in not understanding icons, and a
> >>graphic designer would not change things, I've discussed this
> >>with others in large groups and I'm alone in not understanding
> >>the icon.  Using the TITLE would be wrong,
> >
> > Providing that sort of *additional* information is exactly what TITLE is
> > specified for.
>It is not _addtional information_ it is the information,

No, it's not.  The image is the information.  Presumably, the image conveys 
the message "men at work".

ALT is by definition an alternate means of presenting the message *if the 
image is not present*.  You're either choosing to ignore that basic 
definition, or you don't understand it.

>if someone uses the common men at work road sign, the ALT is "men at 
>work", the title could
>easily be "An example of a UK road sign", I need the ALT to understand it,

If you can't understand the image, then the image is poorly conveying the 
message.  The fact that ALT happens to convey the message in a form you 
understand does not change what the definition of ALT is or what it is 
spec'ed out to do.

>not the title, I need the alternative, I simply do not understand the image

The fact that ALT conveys the message to you does not change what ALT and 
TITLE are designed to do in the spec.

>I do struggle to see why people are so dogmatic on this point, we're in a
>group where everyone is presumably aware of accessiblility, and aware that
>the most important thing is that users are able to get to the content they
>need to understand the document.  I'm a user, my requirements are simple
>compared to a great many of the use cases we see, but I'm still told that my
>User Agent is wrong,

It's not wrong for your needs, clearly.  It is wrong for the spec.

>despite it doing exactly what I require to access a
>well marked up document.  Like the people who need voice browsers don't want
>to force everyone else to use a voice browser, I don't want to force you to
>see ALT on tooltip.

There is clearly a vast difference between forcing a particular UA (or 
class of UA) on users and our argument, so your example has little meaning.

>I just want people to acknowledge that accessibility
>needs are different for every person,

I don't recall saying anything that would contradict that.

>and User Agents should be free to show
>the content any way they please.

In other words, UAs should be free to ignore the very specs you keep 
incorrectly citing to support your position?

>My user agent is not wrong to show ALT on tooltip,

Again, wrong for the spec.  Clearly not wrong for your personal needs.

>neither is yours wrong for not doing so,

Glad you realized that.

>mine's just making the content I need available to me in
>the best way possible, and in following the conventions of my operating

Then by all means, stick to the UA you've chosen.

Bill Mason
Accessible Internet
Received on Wednesday, 5 February 2003 02:59:14 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 20:36:13 UTC