Re: Draft of TAG position on use of unregistered media types in W3C Recommendations

On Jul 21, 2006, at 12:07 PM, noah_mendelsohn@us.ibm.com wrote:
> Bjoern Hoehrmann writes:
> > How about this instead:
>  >
>  >   W3C Working Groups must resolve dependencies on unregistered 
> Internet
>  >   Media Types by directly or indirectly registering these types in 
> the
>  >   IANA media type registry. W3C Technical Reports must not 
> encourage or
>  >   require use or implementation of unregistered media types.

Works for me.

>  > How is Web architecture positively affected by following your 
> policy in-
>  > stead of mine? 
> http://lists.w3.org/Archives/Public/www-tag/2006Jul/0009
>  > discusses how interoperability is negatively affected by your 
> policy.
>
> First of all, thank you for the quick response and the suggestion.
>
> As to the substance, we were contacted in part because particular 
> workgroups were concerned because of a desire to reference particular 
> widely deployed types such as "audio/wav".  While I personally have 
> almost no direct familiarity with the history of debates on this media 
> type, my understanding is that various groups have been involved over 
> the years in so far unsuccessful attempts to register that type.

The registered name seems to be
    audio/vnd.wave;codec=1
per
   http://www.iana.org/assignments/wave-avi-codec-registry
and
   http://www.ietf.org/rfc/rfc2361.txt June 1998

>  So, for good or bad reasons it is in fact both unregistered and in 
> widespread use.  I think that your proposed text would essentially 
> require any W3C group that wanted to produce a Recommendation that 
> exploited such a type to be proactive in reopening debate in IANA (or 
> other appropriate registration body) to get it registered.

Yes, exactly.

>   Speaking for myself and not necessarily for the TAG as a whole, I 
> don't think we should require that in the case where the type is 
> already in widespread use, and where the workgroup in question is just 
> another user.   I believe that the sense of the TAG was to allow some 
> latitude in such cases, and I have tried to capture that in the note.

There's always a certain amount of latitude in the W3C process; I don't 
see much reason to put the latitude into the rules. The rule should be: 
you must get all the relevant media types registered.

By analogy, occasionally people hypothesize namespace names of the form 
http://www.w3.org/xyz
when W3C hasn't issued any namespace nor document called /xyz . I 
haven't seen any of them
in widespread use, but suppose they did... if the IETF had a rule
that using such URIs in IETF specs was OK, I'd be pretty annoyed.

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Friday, 21 July 2006 17:38:12 UTC