W3C home > Mailing lists > Public > public-html@w3.org > July 2007

Re: Improving alt (was handling fallback content for still images)

From: Robert Burns <rob@robburns.com>
Date: Sun, 15 Jul 2007 16:03:45 -0500
Message-Id: <CA881885-5E35-4817-A452-109B874C4CFC@robburns.com>
Cc: public-html@w3.org
To: Sander Tekelenburg <st@isoc.nl>

On Jul 15, 2007, at 2:07 PM, Sander Tekelenburg wrote:

> At 16:45 +0900 UTC, on 2007-07-15, Charles McCathieNevile wrote:
>> On Sun, 15 Jul 2007 06:29:13 +0900, Sander Tekelenburg  
>> <st@isoc.nl> wrote:
>>> What about this then:
>>> - Authors must use no more than n characters as the value of the alt
>>> attribute. For longer alternatives authors must use longdesc.
>> I don't think this makes sense, because as already noted there  
>> isn't some
>> magic number of characters that is useful. Take into account the  
>> fact that
>> some languages require far more characters for the same power of
>> expression and you are walking into the trap of choosing bad  
>> alternatives.
> Yes, very true. But there already *is* a threshold today
> <http://santek.no-ip.org/~st/tests/altlength/>. If your language  
> needs twice
> as many characters as another, today that just means that you're  
> twice as
> likely to have the tooltip disappear before you've read its contents.
> So yes, the problem you point out exists, but I don't see how what  
> I propose
> makes it worse. On the contrary, being more clear about what is  
> meant with
> "short" and "long" makes it easier for authors and UA implementors  
> to auhor
> and implement @alt better.

Also, as I had suggested earlier, we could use a certain number of  
words rather than characters. Unicode takes great pains to define  
word boundaries in all of its scripts. So counting the number of  
words can be done through Unicode algorithms. And I think words is an  
equalizer across scripts in a way that characters just cannot  
accomplish. That is both in expressiveness and in time to read a text  
fragment, words will be a much stronger correlated predictor of how  
much can be said and how long it takes to read.

I also think if the author conformance criteria sets the word limit  
lower than the limit for UA conformance it will help take care of the  
n+1 problem. That is if authors are told to stick to 20 words and UAs  
are told to truncate at 40 words, its only the author's way out of  
bounds that are going to get hit by the limit (or those who push the  
limit to the UA conformance guideline and ignore the author  

Take care,
Received on Sunday, 15 July 2007 21:03:55 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:24 UTC