- From: Mikko Rantalainen <mikko.rantalainen@peda.net>
- Date: Tue, 07 Jun 2005 14:13:56 +0300
- CC: www-html@w3.org, www-html-editor@w3.org
John Foliot - WATS.ca wrote: > Allowing page authors to map explicit keystrokes to elements will render the > access element as useless and flawed as the current accesskey attribute. > [...] > > Please, please, please. Remove the "key" attribute of this new element and > consider returning to the concept of allowing the end user/user agent to map > the user's preferred keystroke combination to roles (pre-defined or other). +1 There isn't enough accelerator keys available that are unused in every browser. Firefox, MSIE or Opera already reserve almost every ALT+anything combination by default. Alt+1 to Alt+9 are used by firefox to directly change to another tab so I cannot use those either as a page author. All unused key ALT-combinations available after testing Firefox, Konqueror, MSIE with Finnish localization and Opera: Alt+q Alt+r Alt+y Alt+u Alt+i Alt+x Alt+c Alt+0 (And at least Alt+q, Alt+r, Alt+c and Alt+x are used by Unix version of Netscape Navigator so don't give any written instructions on the page for these combinations in any case.) *Every* other key combination was already taken by at least one browser. I'm sure that other people can find user agents that use ALT for access key combinator and use the remaining "safe" letters. A non-alphanumeric key cannot be safely used for an access key because international keyboard layouts may not have those easily accessible. That makes grand total of 8 free accelerator keys in the most common case available to content authors. Which one of those would you bind to "search" function (an example of a function that the user might want to use repeatedly) and expect some other author to decide to use the *same* letter? Remember that S is already taken. I think 'role' or 'accesstarget' attribute giving a name for the access point would make much more sense. There's even a remote change that two random authors might select "search" for the function 'role' name by accident even when there weren't any list of defined roles available. -- Mikko
Received on Tuesday, 7 June 2005 11:14:03 UTC