- From: Cynthia Shelly <cyns@microsoft.com>
- Date: Wed, 17 Apr 2002 09:38:54 -0700
- To: "Charles McCathieNevile" <charles@w3.org>, "Phill Jenkins" <pjenkins@us.ibm.com>
- Cc: <w3c-wai-gl@w3.org>
IE has such an action, at least for click. I haven't tried mouseover, but MSDN (http://msdn.microsoft.com) should have documentation on which events can be fired this way. If memory serves, this worked all the way back to IE4. <span tabindex="1" onclick="alert('click');" onkeypress="this.click();">hot span</span> Paste this into an HTML page, tab till the "hot span" has focus, and hit a key. You'll see the 'click' alert. -----Original Message----- From: Charles McCathieNevile [mailto:charles@w3.org] Sent: Wednesday, April 17, 2002 9:13 AM To: Phill Jenkins Cc: w3c-wai-gl@w3.org Subject: Re: onclick and onkeypress scripting technique issue Hi Phill, yes, there is a new issue to deal with here, at the same time as dealing with a legacy issue - IE has linked some events, but not all of them, other developers have implemented the spec as is, and the result is that it is hard for authors. My initial thoughts are that it would be useful if IE had a deliberate "fire the mouse" action that could be keyboard generated, or if the specifications were written to unify the mouse and keyboard models, but having an unspecified half-way solution is not serving us in the long term. On the other hand, it should be acknowledged that they have helped some people in the short term. I recommend that people interested in this question consider the PF working group as a venue for further discussion - there are several groups working on event models including HTML, and SVG. I realise the member-only nature of that group means people will have to join it and agree to the confidentiality requirements, but consider that the work there is valuable. cheers Charles McCN On Wed, 17 Apr 2002, Phill Jenkins wrote: >To: HTML techniques editors > >In the WCAG HTML techniques [1] about making scripts directly accessible, >it says to <quote> provide redundant input mechanisms. In other words, >specify two handlers for the same element such as "onclick" with >"onkeypress" <endquote>. This practice seems to be causing problems >because the IE 5.5 browser (for example) sends two events to the server >when using the keyboard, but only one event when using the mouse. IE maps >the keyboard to some mouse events, such as allowing the user to tab to the >form controls and pressing the enter key instead of using the mouse to >trigger the event, so the second onkeypress event is not necessary. > >1. Is it IE that has a bug? What about Netscape, Opera, Lynx, etc >2. What about the assumption of the User Agents capability and other >software standards? > >508 and UAAG all require the ability to do with the keyboard what can be >done with the mouse. IE and other browsers should allow the user to access >the controls with only the keyboard. Techniques and standards should not >require or put burden on the content author that causes problems, >especially when the burden should and is being handled by the user agent. >Why add on keypress when the device or agent is capable of the mapping? > >So, what is a web developer to do? Following the techniques causes >problems. > >The technique seems to have been established in Feb 1999, or at least that >is when the issue [2] was raised and resolved. But, in my opinion, >resolved incorrectly with today's browser capabilities and today's UAAG >standard [3]. In other words, a new issue should be logged. > >[1] Directly Accessible Scripts >http://www.w3.org/TR/WCAG10-HTML-TECHS/#directly-accessible-scripts >[2] Device independent >http://www.w3.org/WAI/GL/wai-gl-tech-issues.html#dd-events >[3] UAAG requirement >http://www.w3.org/TR/2001/CR-UAAG10-20010912/guidelines.html#gl-device- independence
Received on Wednesday, 17 April 2002 12:39:26 UTC