[whatwg] Comments on Web Forms 2.0 Working Draft

[Christian Schmidt]

Dean Edwards wrote:

> clear as crystal. to support the css3 ":invalid" pseudo class a form 
> control must constantly maintain it's validity state.

Strictly speaking, this is only necessary if there is an :invalid rule 
for the control. If there is no :invalid rule, and control.validity is 
not being queried, and form.validate() is not being called etc., the 
validity state is not imporant.

My idea was to add an onvalidate event that a "stupid" UA would just 
trigger whenever the value changes (i.e. oninput + scripted changes). A 
"smart" UA could be optimized so that the onvalidate event is only 
triggered under the above mentioned circumstances, and perhaps it could 
even skip some calls e.g. if the user is typing quickly in a text field. 
I admit that the non-deterministic behaviour may be a bit too 
fancy/confusing.


Well, I suggest that the spec at least explicitly states that the 
validity state should be constantly maintained.


Christian