Re: [html4all] HTML5 Alternative Text, and Authoring Tools

Matt Morgan-May wrote:
> Visual UAs wouldn't require any new behavior.

I don't believe that's necessarily true.  Visual UAs care about @alt in various 
situations; at the very least any time images are disabled, if the image data 
can't be loaded for some reason, or if the image is in a graphics format the UA 
doesn't support.

Granted, these are somewhat corner cases for visual UAs in terms of how often 
our users hit them, but nevertheless they are ones we try to take pretty seriously.

> And not whether that is intentional, for which I'm proposing @noalt, or
> accidental, in which case AT would need to attempt to repair the missing
> data. The implications for these two scenarios is very different at the
> UA/AT level. 

I'm finding this argument somewhat compelling, as our armchair Aristotelian 
arguments have gone, to be honest.  So if I understand correctly, the proposal 
here is that from a validator's standpoint, precisely one of @alt and @noalt 
must be present on every <html:img>.  Is that correct?

I think I saw an argument go by that claimed that this is equivalent to having 
lack of @alt imply @noalt, but that claim does assume that everyone is producing 
valid documents...

The other concern I can see here is that authors would sprinkle the @noalt pixie 
dust the way some claim the sprinkle the alt="" or bogus alt value pixie dust. 
This part could use a bit of validator usability study.

> There's nothing wrong with getting content providers to show their colors,
> IMO.

True, as long as it doesn't interfere with actual accessibility improvements 
(both in the sense of resource diversion and in the sense of color-showing 
involving changes that actually reduce accessibility).  Evaluating whether such 
interference occurs can be nontrivial, of course.


Received on Wednesday, 14 May 2008 21:08:17 UTC