W3C home > Mailing lists > Public > public-web-notification@w3.org > December 2013

Re: RfC: LCWD of Web Notifications; deadline October 24

From: Marcos Caceres <w3c@marcosc.com>
Date: Mon, 2 Dec 2013 12:03:38 +1000
To: Andrew Wilson <atwilson@google.com>, Anne van Kesteren <avankesteren@mozilla.com>
Cc: Jon Lee <jonlee@apple.com>, public-web-notification <public-web-notification@w3.org>, Arthur Barstow <art.barstow@nokia.com>
Message-ID: <499FE46CB07E455785958752035D82B8@marcosc.com>


On Friday, November 29, 2013 at 2:14 PM, Andrew Wilson wrote:

>  
> On Fri, Nov 29, 2013 at 2:59 PM, Marcos Caceres <w3c@marcosc.com (mailto:w3c@marcosc.com)> wrote:
> >  
> >  
> > On Friday, 29 November 2013 at 08:51, Andrew Wilson wrote:
> >  
> > > I don't think the W3C spec can or should claim primacy over the WHATWG specification - they are just different (albeit related) entities. In fact, I'd argue that it's not the role of the W3C spec to highlight differences between itself and any other specs out there, nor should it be instructing implementors to ignore other specifications.
> >  
> > But this one just copy/pastes from the other one, no? That's completely different than saying "this spec is different from some other spec".
>  
> The long and convoluted history of the Web Notifications spec is too long to go into here.  
Sorry, it’s not my intention to open up any old wounds. I don’t know the history.    
> Suffice it to say that "one copy/pastes from the other" is probably an oversimplification. As is saying that one spec is somehow canonical while the other is a fork. If you're looking for guidance, I will note that Chrome has implemented the W3C specification, but has not implemented any of the additional APIs (like Notification.get()) in the WHATWG spec.
>  


Knowing which spec Chrome is working from is very useful information; thank you! I guess it’s the same with WebKit? Any idea about Microsoft?    
  
>  
>  
> > If one spec is going to copy/paste from another one, then it should be clear that they are exactly the same bar the boilerplate (and are being synchronised and how often). If the spec has modified the source spec (the WHATWG one), then it should be clear in what way they differ.
>  
>  
> I do not believe that there is any intent on keeping this spec synchronized with the WHATWG document. Jon can correct me if I'm wrong here.

If we know which one implementers are working from, then we should throw out one of them.  
  
>  
> > I don't know which spec to review otherwise. Or why I should bother paying attention to this one (not as a judgement, but from an implementation perspective). Also, what happens if the WHATWG changes once you enter LC or CR in a substantive way?
>  
>  
> You'd have to talk to Anne (the former W3C spec editor and current WHATWG spec editor) about his goals for that spec.
Anne? Can you please clarify?   
> Given that there are already multiple implementations of the W3C LCWD, I'd say it's unlikely that he would specify anything that is not backwards compatible with this document.
>  


Anne? Which one are we following at Moz?  
  
>  
> > > In this specific case, the WHATWG spec is a living document, so any attempt to enumerate differences would quickly become stale.
> >  
> > Does that mean that you are planning to freeze this spec? Did you copy and paste from a particular date in time and fork from there? Is this a fork?
> >  
>  
>  
> Not sure what "freeze this spec" means,
It means that the feature set would be frozen - and that no more copy/pasting from the WHATWG spec would happen at that point in the current W3C spec. For example, the “.get()” method or other WHATWG editions would not be added during LC and obviously not during CR.   
> but I think that the point of having a LCWD is to have a canonical revision that implementors can all reference. Having that document contain some list of differences relative to another document that can change at any arbitrary time seems like it's contrary to the point of having an LCWD.
>  

It doesn’t have to be. If the document states:

“The bulk of this text was copied on the [date] from the WHATWG spec - and it’s accurate up to this commit [git commit hash]” - then one would know when the forking occurred.  

At some point, a V2 version of the spec will have to happen to include any new methods Anne is adding on the WHATWG side.  

If the bulk of implementers are working off the W3C of, then the WHATWG version should layer it’s extensions on top of the W3C one, no?  

It sounds like the two specs are diverging - and this makes me worry.  

Kind regards,
Marcos  
Received on Monday, 2 December 2013 02:04:07 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 19:53:14 UTC