[whatwg] LABEL and radio/checkbox onclick

[Matthew Raymond]

Jim Ley wrote:
>>   Because otherwise part of the web page UI may behave in
>>unpredictable ways, causing an author's custom UI work to fail on
>>specific platforms.
> 
> So that's a good reason why not to create your own custom UI
[Stuff about old versions of Mozilla conflicting with a feature in 
X-Mouse snipped.]

    People are going to create custom UI regardless of what you think. 
They do it all the time. Furthermore, in most cases they design it for a 
specific platform: Windows. This means that if you put too many UI 
restrictions on a browser, those browsers will have problems dealing 
with UI designed with a different OS in mind.

    That's why you need some level of behavior and UI description in a 
spec. If you don't have that, people will simple follow the behavior and 
UI of the most common browser on the most common platform.

    That's not to say we can't give the vendors choices of how to 
implement features, behaviors and UI. It just means that if we are too 
vague when it counts, people will turn to specific platforms and vendors 
as their guide.

>>   Why can't the expect the phone and the PC to behave the same when
>>rendering the same markup and styling?
> 
> That's not what we were discussing, we're discussing interaction
> behaviour, and obviously a phone with no pointing device is going to
> be different to a system with a mouse.

    What's your point? The HTML specification talks about focus, not 
pointers. What does it matter if there is no mouse, so long as there is 
a way to give a control focus.

>>   By this reasoning, specifications should use the lowest common
>>denominator of OS behavior only, or not specify behavior at all. 
> 
> The latter, just like HTML shouldn't require a rendering for its elements.

    What's the point of HTML if you can't render it?

>>You want clicking the label to be
>>treated as a click on the checkbox because the checkbox is such a small
>>target to hit with your mouse that it becomes a usability issue.
> 
> But that doesn't need to be specified, as you say it's a usability
> issue, but that just means no UA developer is going to do it, as their
> UA won't be usable.

    We can't always rely on vendors to create proper UI. Internet 
Explorer doesn't properly implement label passing to an implicit radio 
button or checkbox control, in violation of both W3C spec and OS 
conventions. How can we ask for proper UI from other vendors if it's not 
even in the spec?