[whatwg] Introducing new elements is expensive

[Matthew Raymond]

Ian Hickson wrote:
> On Sat, 12 Mar 2005, Matthew Raymond wrote:
>>That is not the case with <output> in a legacy user agent. Any HTML 
>>contents will display as HTML in a legacy UA. Yet if we assume the 
>><textarea> model for <output>, WF2 UAs will display the underlying 
>>markup as text instead of rendering it. So, right off the bat, we have a 
>>difference in how the contents of <output> are rendered between legacy 
>>and WF2 user agents.
> 
> Elements inside <output> are rendered as elements, with all that that 
> implies; there is no special parsing.

    The contents of <output> are the initial value, yet they are not parsed.

    /me scratches his head.

>>Since we can't avoid a difference in rendering unless we artificially 
>>enforce a no-markup-inside-<output> rule, what does it hurt to simply 
>>have a |value| attribute to set the .defaultvalue directly?
> 
> The two things are separate. What's the use case for value=""?

    Well, for one, in situations where you want to use <input readonly> 
as a fallback container for a calculated value.