Re: Alternate proposal for ISSUE-30 longdesc

On Mon, 22 Feb 2010, Jonas Sicking wrote:
> On Mon, Feb 22, 2010 at 7:05 AM, Maciej Stachowiak <> wrote:
> > Let me turn it around. What do you think is the value of making 
> > longdesc nonconforming instead of conforming with a warning? (I 
> > mentioned what I thought was the downside of conforming-with-a-warning 
> > compared to nonconforming in my Impact section, but I am curious what 
> > you think.)
> If we have determined that aria-describedby is a better attribute from 
> an accessibility point of view, I think we should put full force behind 
> that message and deprecate longdesc rather then by being wishy-washy. 
> This seems like the best way to help accessibility.
> Additionally, from a language lawyer point of view, the whole point of 
> conforming vs. non-conforming to me is that we, as spec authors, 
> recommend that people use things that are conforming, and recommend that 
> people don't use things that are non-conforming. That to me is the sole 
> purpose of the term "conforming". (Note that implementation requirements 
> are unaffected by conformity).
> So the concept of "conforming but with a warning" seems self 
> contradicting.

This is why I prefer the old mechanism of "down-played errors" -- things 
that are non-conforming, but where the validator is allowed to gloss over 
the error and just say "oh and by the way there were these other things 
that were wrong, but these are no big deal". It's still clearly non- 
conforming, but we recognise that there are some errors that just aren't 
as critical to fix (especially if you're transitioning from valid HTML4 
Strict documents) as other errors.

Ian Hickson               U+1047E                )\._.,--....,'``.    fL       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Monday, 22 February 2010 19:52:42 UTC