[whatwg] <input type="text" accept="">

[Matthew Raymond]

Alexey Feldgendler wrote:
> On Sun, 11 Jun 2006 23:29:39 +0700, Anne van Kesteren  
>> I guess it looks like it would fit in CSS because the functionality is  
>> not strictly needed, but I'm unsure if it's really just presentation...
> 
> Ok, it's not just presentation. It's about behavior, too. But I don't  
> think that it's wrong to use CSS for behavior. In fact, IE already does  
> so, and I think it's one of IE's strengths.

   IE supports HTML Controls (HTC), and from what I hear, the
implementation of this proprietary standard has so many problems it
should be considered one of IE's weaknesses, not one of its strengths.
By comparison, XBL is so powerful that Dean Edwards used it to implement
HTC support in Mozilla.

   IE supports HTCs via a CSS property called "behavior", which is
pretty much the same as -moz-binding. (Only you can actually tell that
the latter is a vendor extension.) In both cases, we're talking about
the use of a single property in CSS, with the rest of the work being
done in markup (either XBL or HTC). Note also that XBL can included
inside XML compound documents and can bind to elements without CSS
properties, which means that XBL can add behavior to markup without CSS.

   While behavioral styling has its uses, support for behavior in CSS
should not exceed what behavior/-moz-binding does. Going further than
that doesn't just corrupt CSS, it brings nothing new to the table. In
the future, behavior not linked to a specific style sheet will hopefully
be done via XHTML+XBL, with a "binding" property and XBL handling the rest.