Re: summary attribute compromise proposal

Maciej Stachowiak wrote:
> ...
> I believe existing AT will read the caption, text next to the table, or 
> text associated via aria-describedby.
> ...

Even if it's set to hidden via CSS?

>> In my opinion, warnings make only sense if there's a way for the 
>> author to silence them. For instance, in Java, for deprecated methods 
>> the JavaDoc usually specifies an alternative. And for many other 
>> warnings, there are pragmas to point out "I mean it".
> 
> Do you think an out-of-band way to silence the warnings (such as a 
> validator option to turn warnings on and off) would be sufficient to 
> meet this need? (Not necessarily proposing this - I'm just trying to 
> understand the scope of your disagreement.)
> ...

No, I think it would need to be in-band.

> ...

BR, Julian