[whatwg] Accesskey in Web Forms 2

Matthew Thomas mpt at myrealbox.com
Sat Jul 24 08:06:01 PDT 2004

On 25 Jul, 2004, at 12:04 AM, Jim Ley wrote:
> On Sat, 24 Jul 2004 14:14:30 +0300, Kai Hendry <hendry at cs.helsinki.fi>  
> wrote:
>> 1) Users can not easily tell what the accesskeys are.
> Their UA should do a better job of informing them.

Agreed. (For example, four years ago I requested for Mozilla: "Where an  
HTML element has an ACCESSKEY, if a character (case-insensitively)  
equivalent to the accesskey is present in the element's label, that  
character should be underlined by default."  

>> 2) Accesskeys can conflict with the UA binds.
> Their UA should do a better job of picking a key combination to
> activate them that doesn't interfere with their default UA binds.
> There's lots of possible modifier keys available to a UA -

Disagreed. Perhaps you could name some of these "lots of possible  
modifer keys available". Four years ago, I couldn't find any on  
Windows, and only Control on Mac OS.  
<http://bugzilla.mozilla.org/show_bug.cgi?id=51940#c4> Since then, the  
situation has only become worse. (To wit, Mac OS X has appropriated the  
Control key for emacs keybindings in text fields. This makes its  
behavior unpredictable for HTML accesskeys, since the thing you are  
trying to move focus away from may well be a text/password field.)

> why override ALT.

Because that's what the native GUI uses, so if users expect it to be  
anything, they expect it to be Alt. (Which is unfortunate, since UAs  
are already using Alt for other things.)

> Yes AccessKey isn't that well specified, but I think it can be fixed
> rather than completely removed.
> ...

If mass-market UAs are reliably browbeaten by cries of "accessibility!"  
into implementing accesskey when they shouldn't (i.e. when all  
available modifier keys are already used by the UA or the operating  
environment), perhaps the world would be better off if the attribute  
was removed from HTML.

However I doubt WhatWG will be doing that, since Ian has said "A  
document that conforms to HTML4 Strict will always conform to any  
relevant spec created by WHATWG"  
001330.html>. Removing accesskey from HTML5 would break this dictum by  
making some HTML4-Strict-conformant documents HTML5-nonconformant.

What the Web Forms spec *could* do is recommend that UAs ignore the  
accesskey attribute in those environments where implementing it would  
conflict with environment conventions (in the same way as it could  
recommend that UAs follow environment conventions for <label>). If  
necessary, the spec could even include a statement discouraging authors  
from using the attribute.

Matthew Thomas

More information about the whatwg mailing list