[whatwg] Re: several messages

[Matthew Raymond]

James Graham wrote:
> Matthew Raymond wrote:
>>   When using the inheritance feature of <idate>, incompatibility 
>>isn't the default either, and the only situation in which you can't 
>>use inheritance is when the first child control doesn't submit a 
>>complete date. You're arguing a "Rogue Webmaster" scenario.
> 
> To be clear, my adaptation of this model did not include any such 
> inheritance (for implementation simplicity). Having said that I don't 
> think that requiring authors to explicitly provide fallback content is 
> such a bad thing (at least, not worse than the limited fallback options 
> offered by input). But more on that later.

    I could be convinced to drop inheritance if it's shown that it 
creates significant implementation problems, but unless someone can 
clearly communicate the problems involved, I still believe any potential 
feature removal should occur during the implementation phase of the spec.

    On a slightly unrelated note, I just realized a short time ago that 
the Microsoft sabotage scenario could happen just as easily if <idate>, 
or any part of it, were left out of the spec as it would be if it were 
included as-is. All Microsoft would have to do is implement flawed 
non-inheritance support (to mess up my version of <idate>), or implement 
my version in its entirety for either Ian's position or yours. So I 
don't think manipulating the markup is going to keep Microsoft from 
screwing us if it so desires...