Re: WOFF2 file extension and media type

Hello Vladimir,

Monday, June 30, 2014, 3:50:24 PM, you wrote:

> Thank you for your prompt response. I agree that file extension
> “.woff2” would be ideal so, if there are no objections from anyone
> and no limitations on the  string length from any existing software
> vendors (anyone?) – let’s finalize this as our recommendation and add it to the spec.

Great.
>  
> With regard to media type – I am a bit hesitant to recommend
> “fonts/woff2”. The top level “fonts” type isn’t registered yet and I
> am not sure how long would  it take for the registration application
> to be processed by IANA.

Agreed; surprised to see it in use.


> The subtype in the “application” tree is
> definitely an easier choice since the registration is going to be
> quick and the positive outcome is almost guaranteed. This however
> brings up another  question – how critical is it to have the media
> type registered?

I have seen developers refuse to use a type because it wasn't
registered,or (as in this case) to make up their own types. I have
also seen inaccurate, guesswork types added to Wikipedia to cover a
gap in registration (and then persist or get re-added).

To allow parallel deployment with WOFF 1.0, there needs to be a new
media type and a new "format string" used in @font-face.

Also, W3C requires specifications that define new media types to
submit them to IANA for review at or before W3C Last Call.

WOFF 1.0 uses application/font-woff

I think therefore that it is crucial to add a Media Type appendix to
the WOFF 2.0 specification. I have one prepared. if there are no
pending edits, I will add it. if there are pending edits than (if you
are still having CVS issues, Vlad) please send me the latest document
and I will check it in, then add the appendix and check that in.


>  Who is going to benefit from it? I am surprised
> that Google folks unilaterally decided to use “fonts/woff2” type
> that has no legs to stand on.

I was surprised to see that too.

> I always assumed that media types have
> to be registered  and universally understood in order to be a useful
> utility, but if one can just invent a new type and start using it
> without registering and making it known to the rest of the world –
> how important this media type business really is? (I am asking this
> questions  because after many years of discussions about it I still
> do not understand the value of media type and whether putting
> significant efforts into the registration procedure would pay off in pragmatic sense.)

Pragmatically, the most important thing is that one exists and that
everyone uses it. (The pre-emptive use of font/* ahead of a new
top-level registration works against that, by splitting useage).

The actual string does not in practice make much difference beyond
aesthetics.

It is also increasingly difficult to change once something is deployed
(for example, although VRML is not much used anymore, the official
model/vrml is not used and is not interoperable while the old,
"temporary" x-world/x-vrml is in practice what everyone uses). But the
use of x- prefixes means that this type cannot be standardized. The
use of an unregistered top level type would also preclude or at least
defer standardization (until the top level type was standardized).

So please let us define, standardize and *use* application/font-woff2
and not muddy the waters with any other types.

I will also ensure the @font-face format string woff2 gets added to
CSS3 Fonts, as already used in the example in our evaluation report
http://www.w3.org/TR/WOFF20ER/#deploy

-- 
Best regards,
 Chris                            mailto:chris@w3.org

Received on Monday, 30 June 2014 15:57:56 UTC