Re: [XHR] Last Call comment on about dependencies

Hi Anne,

Thanks for your reply. (We are assuming that this is not a formal reply  
 from the webapps WG.)

>> The XHTML 2 working group discussed the XHR draft at a recent  
>> teleconference, and I was asked to send in a brief comment. Basically,  
>> the XHTML 2 Working Group is concerned that the draft appears to have a  
>> dependency on HTML5.  On closer inspection, it is not clear whether  
>> this dependency is completely necessary.  Further, linking the spec to  
>> HTML5 will delay its deployment and incorporation into other languages  
>> that have a vested interest in portable scripting (e.g. XHTML 1, XHTML  
>> 2, XForms).
>
> The concepts defined in HTML5 are important for getting interoperable  
> implementations of XMLHttpRequest. I don't think deployment is  
> necessarily an issue as XMLHttpRequest is already deployed. Now we just  
> need to make sure the various user agents get in line with respect to  
> the behavior they each have.
>
> It's not entirely clear to me why XMLHttpRequest needs to be  
> incorperated into a language. In fact, it was incorperated in HTML5 at  
> some point and we splitted it out. (Given the amount of work this cost  
> me I'm still not sure whether it was worth the cost, but it has been an  
> interesting exercise nonetheless.)
>
>
>> Finally, it appears that the dependecy is slightly backwards, since the  
>> requirement is that HTML5's document "Window" object support the  
>> XMLHttpRequest interface.
>
> Actually, no. The requirement is that objects implementing the HTML5  
> Window _interface_ also support the XMLHttpRequest constructor.  
> Furthermore, the definition of HTML5 Window is important here in case of  
> URI resolution in cross-frame scenarios.
>
> Also, that is not the sole dependency the XMLHttpRequest specification  
> has on HTML5.
>
>
>> Our request is that this dependency be removed (or that the connection  
>> be made informative instead of normative) so that all interested  
>> constituents can take advantage of this important interface as soon as  
>> possible.
>
> I don't think this is possible. Feel free to go through the  
> public-webapi mailing list archives to find more detailed discussion on  
> this subject if you feel the above is not sufficient:

There seem to be several options:

1. XMLHttpRequest is irrevocably bound to HTML5.
    If that is the case then there seems to be no reason to develop this  
spec outside of the HTML5 WG, or indeed for developing as a separate spec.
2. XMLHttpRequest is host neutral, and therefore can be used in different  
environments.
    If that is the case, and it would seem preferable since there are  
several other technologies that are able to use this, then it seems good  
to make it as widely adoptable as possible. It seems like there are two  
ways to do this:
    a. copy the restrictions due to HTML5 into this document, so that it is  
free-standing
    b. remove the restrictions due to HTML5, and ensure that they are added  
to that spec, and let languages that use it specify the necessary  
restrictions needed to make it work in that environment.

Which of these do you think best apply to XMLHttpRequest?

There seem to be several technologies in W3C that could use  
XMLHttpRequest; SMIL and XForms come readily to mind. Would you be able to  
enumerate what it is in XMLHttpRequest that is so bound to HTML5?

Thanks!

Best wishes,

Steven Pemberton

Received on Wednesday, 25 June 2008 13:48:35 UTC