Re: referencing W3C or WhatWG specs (was: [webauthn] new commits pushed by equalsJeffH

On 9/14/16, 9:07 AM, "Vijay Bharadwaj" <vijaybh@microsoft.com> wrote:
>In the PR I sent out last night
>(https://github.com/w3c/webauthn/pull/198) I took the second approach by
>pointing to HTML 5.1 which is in CR. This seems like a
>best-of-both-worlds compromise since HTML 5.1 appears to address all the
>things we care about (in many cases by pointing through to WHATWG specs,
>but thatıs not an issue for us to worry about).

Though, note this new issue from @bzbarsky..

  The W3C HTML spec is broken, and probably shouldn't be referenced
      
  https://github.com/w3c/webauthn/issues/257


..where he notes that in at least one case where we reference into html51,
the latter spec is broken.



> 
>From: Hodges, Jeff [mailto:jeff.hodges@paypal.com]
>Sent: Wednesday, September 14, 2016 8:52 AM
>To: Wendy Seltzer <wseltzer@w3.org>
>Cc: public-webauthn@w3.org
>Subject: Re: referencing W3C or WhatWG specs (was: [webauthn] new commits
>pushed by equalsJeffH
>
>
> 
>On 9/14/16, 7:39 AM, "Wendy Seltzer" <wseltzer@w3.org> wrote:
>
>
>>On 09/13/2016 10:45 PM, =JeffH via GitHub wrote:
>>
>>>The following commits were just pushed by equalsJeffH to
>>>
>>>
>>>https://github.com/w3c/webauthn:
>>>
>>>* ref whatwg HTML spec for origin & Navigator. Fixes #160.
>>>
>> 
>>
>>Are these references not available in W3C HTML? I didn't read that into
>>
>>#160.
>>
>> 
>>
>>https://w3c.github.io/html/browsers.html#section-origin
>>
>>https://w3c.github.io/html/webappapis.html#the-navigator-object
>>
>> 
>>
>>We'll face normative reference[1] and consistency questions if we use
>>
>>the WHATWG reference.
>>
>>
>>
> 
>
>yep, tho the W3C and WhatWG need to actually address this overall spec
>dichotomy mess. Left unresolved it puts all of us in tough spot.  See..
>
>Figure out how to manage normative dependencies to WHATWG HTML for
>features not in W3C HTML for CR/REC
>https://github.com/w3c/webappsec-secure-contexts/issues/9 ..especially..
>
> 
>
>[0] 
>https://github.com/w3c/webappsec-secure-contexts/issues/9#issuecomment-233
>633338 
>
> 
>
> 
>
>So it seem that editors of W3C specs depedendent on
>HTML/URL/DOM/FETCH/ENCODING/etc can take one of two approaches at this
>iterating-on-working-draft time:
>
> 
>
>1.  use all whatwg references  (eg for things in
>HTML/URL/DOM/FETCH/ENCODING), then at "transition to CR" time, assess all
>such refs and for those which correctly
> exist in extant W3C specs, update them to point to the W3C specs.
>
> 
>
>2. attempt to use all W3C references as we go along, but for those that
>do not exist or are no longer accurate (eg the W3C spec is stale relative
>to current implementations),
> use the appropriate WhatWG reference (those specs are claimed to be the
>implemented bleeding edge (?)).  At "transition to CR" time, assess all
>refs and update as appropriate.
>
> 
>
>the commit you're pointing to is taking approach (1); for me, it seems
>the easier path for the time being while we're iterating on a working
>draft.   i.e. in trying
> to discuss Anne's feedback with him, he's only referring to whatwg
>specs. 
>
> 
>
>Though, we as a working group can decide which approach ( (1) or (2) ) we
>wish to take, and I'll update it per the decision.  Perhaps we should add
>this to our
> discussion topics next week in Lisbon.  In any case, we should certainly
>be having Lisbon side-discussions wrt @mikewest's observations in [0].
>
> 
>
>hth, 
>
> 
>
>=JeffH
>
> 
>
> 
>
>> 
>>
>>--Wendy
>>
>>[1]
>>https://www.w3.org/2013/09/normative-references
>>
>> 
>>
>>>   by JeffH
>>>
>>>https://github.com/w3c/webauthn/commit/e8172c4d2eea64bf60e0f53d82cb96fa0
>>>81db22e
>>>
>> 
>>
>>
>>
>
>
>

Received on Monday, 7 November 2016 00:14:31 UTC