Re: getting rid of callers

On Tue, Oct 12, 2010 at 10:47 PM, Maciej Stachowiak <mjs@apple.com> wrote:
>
> On Oct 12, 2010, at 9:14 PM, Cameron McCormack wrote:
>
>> Various HTML*Collection interfaces and the HTMLFormElement interface
>> define callers.  Callers seem unpopular with many people, so I am
>> wondering how many can actually be safely removed from the spec.
>>
>> Here are some tests I ran to see where they’re implemented:
>>
>>  http://people.mozilla.com/~cmccormack/tests/callers.html
>>  http://people.mozilla.com/~cmccormack/tests/callers-quirks.html
>>
>> The results are here:
>>
>>  http://people.mozilla.com/~cmccormack/tests/callers-results.html
>>
>> (I left out the HTMLPropertyCollection ones since nobody implements that
>> yet.)
>>
>> So of all the callers, Firefox has been getting away with only
>> implementing document.all("blah"), and only for quirks mode.
>>
>> For HTMLFormElement, it seems only IE implements them.  Are they really
>> necessary to have in the spec?
>>
>> I don’t have any data on any of this.  Does anyone else?  Any browser
>> vendors willing to drop any of these callers? :)
>
> From your results, it seems like Firefox is the odd man out. I think the case against would have to be pretty compelling for every other browser to change to match Firefox.
>
> I realize some people find custom call behavior distasteful, but it's not really clear to me what the practical problem is with it.

While I'm not a fan of call behavior in general, I don't think that is
the big issue here. I think we generally should stay away from adding
*any* feature to the platform unless it is either a well designed and
useful feature, or a feature that is needed for web compat. So far
neither has been shown to be the case here.

> Also, I find quirks-mode-only DOM behaviors more distasteful than custom call behavior, so if document.all needs it at all, I'd rather do it everywhere than limit to quirks mode.

Yup, I agree.

/ Jonas

Received on Wednesday, 13 October 2010 06:27:25 UTC