Re: summary="" in HTML5 ISSUE-32

Hi Joshue,

Thanks for commenting on these alternatives. I'm not in a position to  
argue for one of these as the best way to do things, or that any of  
them is better than summary="".  But I hope we can all agree that they  
are at least plausible ideas; even if they ultimately don't work, we  
should assume anyone advocating them is doing so in good faith. One  
minor comment below:

On Feb 25, 2009, at 2:51 PM, Joshue O Connor wrote:

> Maciej Stachowiak wrote:
>> Many alternatives to summary have been proposed. At least the  
>> following
>> techniques have been suggested as ways to provide additional  
>> information
>> about tables:
>>
>> 1) Use the table caption.
>
> Useful as a terse descriptor only.
>
>> 2) Use a separate piece of explanatory text above or below the table.
>
> The explanatory text is usually contained within the @summary  
> contents.
> If it is needed for sighted users for whatever reason it could be
> exposed via scripting or UA behaviour could be changed in the future  
> to
> support exposing it if required.
>
>> 3) Use the title attribute on the caption.
>
> Not sufficient. I don't think I have ever come across this technique
> being used and would not wish to start to advocate for its support  
> when
> there is an existing attribute that serves the purpose adequately.
>
>> 4) Use a details element inside the caption.
>
> Again, I can't see this flying at all and why do this when there is an
> existing attribute that serves the purpose adequately?
>
>> 5) Use a separate piece of explanatory text above or below the table,
>> using media-specific CSS rules to hide it from visual users.
>
> Why do this when there is an existing attribute that serves the  
> purpose
> adequately?
>
>> 6) Variants of 4 and 5 using ARIA to associate the information with  
>> the
>> table.
>
> Yes, this method could fly but I can't see why you would want to do  
> this
> when there is an existing attribute that serves the purpose  
> adequately?
> There is an argument for promoting the use of ARIA development
> techniques by applying it in instances like this but again, if it  
> ain't
> broke don't fix it right?
>
>> There has also been discussion about whether different techniques  
>> might
>> be appropriate to different kinds of additional information:
>>
>> A) Descriptions of the table's data layout, making it easier to  
>> navigate
>> the table.
>
> @summary already does this.
>
>> B) Conclusions that summarize the data in the table, so one could  
>> skip
>> the table entirely if not interested in the details.
>
> @summary already does this.
>
>> C) Additional information not found in the table at all, but  
>> relating to
>> its contents.
>
> @summary could do this.

I don't think anyone disputes that the summary attribute could convey  
this kind of information. The disagreement is over whether it is  
better to use summary="", or a mechanism that would also make the  
information available to sighted users. Because, for example, you  
could also say that a <p> element above the table "already does this"  
but that's not very helpful to making the call on which way is better.  
Category C especially seems to me like something it would be wrong to  
put in summary - if the information truly can't be found anywhere  
else, then it seems to violate the ideas of equivalent content and  
universal access to hide it from sighted users.

>
>> D) Indications that this table is a layout table.
>
> Use of a null @summary will render the table transparent without
> bothering the user with info like @summary="This is a layout table and
> if you are accessing this data you are probably using a screen reader
> which means that is would probably be better if this information was
> actually transparent as its is kinda useless to you". [1]
>
>> Perhaps there are other kinds of information that might be conveyed  
>> in
>> the summary attribute and that are useful for use of tables with
>> assistive technologies.
>
> Yes, this is only limited to the descriptive abilities of the author.
>
>
> [1] http://www.w3.org/TR/WCAG20-TECHS/H73.html
>
>
>

Received on Wednesday, 25 February 2009 23:32:44 UTC