W3C home > Mailing lists > Public > public-html@w3.org > August 2009

Re: summary attribute compromise proposal

From: Maciej Stachowiak <mjs@apple.com>
Date: Tue, 04 Aug 2009 13:02:50 -0700
Cc: HTML WG <public-html@w3.org>
Message-id: <10BF7409-E72C-4562-8599-12747FCC06D8@apple.com>
To: Julian Reschke <julian.reschke@gmx.de>

On Aug 4, 2009, at 12:44 PM, Julian Reschke wrote:

> Maciej Stachowiak wrote:
>> ...
>> To be really clear, the purpose of the warning is to give authors  
>> the chance to *consider* other approaches, not to outright rule out  
>> summary. It wouldn't say "don't use summary", it would say  
>> something like "if you're using summary, you may want to consider  
>> these issues and these alternate approaches...". So an author could  
>> see the warning and decide they have good reason to use summary=""  
>> anyway.
>> ...
>
> So a good reason would be that @summary is used in exactly the way  
> it was specified?
>
> "This attribute provides a summary of the table's purpose and  
> structure for user agents rendering to non-visual media such as  
> speech and Braille." (HTML4)
>
> (potentially clarified)

If the author has information that's useful in non-visual media, and  
is confident that the information would be unhelpful and distracting  
in visual media, (to people of normal ability and to member of other  
handicap groups besides the blind), then summary="" may be an  
appropriate technique. There may be other edge case reasons for the  
info to be inappropriate in visual media, such as Roy's example, where  
the goal is to give a faithful visual rendering of an existing  
document, while also providing accessibility for the blind.

I think the HTML4 statement as given, while not exactly wrong, gives a  
skewed perspective. It has a built-in bias of providing summary info  
only in non-visual media, and doesn't give due consideration to cases  
where such info may be useful in visual media as well. HTML5 should  
carry a bias towards providing summary info to everyone in all media,  
while allowing for reasonable exceptions. So I don't think I could be  
on board with just replicating the HTML4 guidance.

> The reason why I ask is the fact that authors do strange things to  
> get rid of warnings, including doing wrong things (like blindly  
> adding @alt="")

I think the validator should be really clear that it's not necessary  
to get rid of all warnings. That's why they are warnings, not errors.  
I agree that we shouldn't create a situation like blindly adding  
@alt="" for badge-seeking purposes.

>
>> I think this is appropriate, because HTML4 did not have any other  
>> recommended techniques for table descriptions, so the warning will  
>> give authors a good chance to consider other approaches.
>> It's also similar in spirit to validator.nu's "image report"  
>> feature, which will help you ensure that your use of images is  
>> accessible but without commanding one specific way to do it.
>> ...
>
> I agree it's good to make sure @summary is used for what it's there;  
> I'm not yet convinced that an unconditional validator warning is the  
> right way to get there, though.

Do you think it's acceptable as part of a compromise, even if you're  
not sure it's ideal?

Regards,
Maciej
Received on Tuesday, 4 August 2009 20:03:36 UTC

This archive was generated by hypermail 2.3.1 : Friday, 10 October 2014 16:24:50 UTC