Re: Anything to discuss during our telcon tomorrow?

Named instances can also specify a PostScript name, which would allow selection even within a collection.

> On Dec 14, 2016, at 1:51 PM, Jonathan Kew <jfkthame@gmail.com> wrote:
> 
> On 14/12/2016 21:01, Chris Lilley wrote:
>> 
>> 
>> On 2016-12-14 15:23, Roderick Sheeter wrote:
>>> Cool. Hey, would #postscriptname work for any postscript name? In
>>> particular, for accessing a named instance in a variation font?
>> Huh, interesting. I hadn't thought about that one!
>> 
>> I think the plan is to access anywhere along an axis by putting the axis
>> values on the @font-face descriptors. So named instances are not privileged.
> 
> I was going to say it should: named instances should be available by name. But there's a complication with that. In principle, you could have a collection containing multiple variation fonts, and multiple named instances within each of them. Those names might not be unique. So it'd be perfectly reasonable (I think) to have:
> 
>  MyFamily.ttc
>    face: MyFamily-Upright
>      instances: Light, Regular, Bold, Black
>    face: MyFamily-Italic
>      instances: Light, Regular, Bold, Black
> 
> as instance names are meaningful only within a single (variation) face.
> 
> Therefore, while I think it is useful to provide a means to access named instances (without having to discover and specify their actual variation coordinates), this needs to be distinct from the use of the PostScript name to select a face within a collection.
> 
>> 
>> Would non-variation-aware processors be able to get at named instances?
>> (for TrueType glyhs - clearly they would not be able to for CFF2)?
>> 
> 
> I wouldn't expect so. A named instance is just a name for a specific set of variation-axis values; but to apply those values, the rasterizer needs to know how to process variations.
> 
>>> 
>>> On Wed, Dec 14, 2016 at 11:40 AM, Chris Lilley <chris@w3.org <mailto:chris@w3.org>
>>> <mailto:chris@w3.org <mailto:chris@w3.org>>> wrote:
>>> 
>>> 
>>> 
>>>    On 2016-12-13 23:14, Levantovsky, Vladimir wrote:
>>>> 
>>>>    On a related subject – there have been updates on the top-level
>>>>    font media type registration. Chris Lilley has been busy at work
>>>>    (thank you Chris!) addressing some of the issues reported by the
>>>>    IETF-assigned reviewer,
>>>> 
>>>    all of them, I hope :)
>>>> 
>>>>    and the new version of the document has been created as a result.
>>>>    Please review and send you comments, if any. The details on the
>>>>    document progression can be seen at
>>>>     https://datatracker.ietf.org/doc/draft-ietf-justfont-toplevel/ <https://datatracker.ietf.org/doc/draft-ietf-justfont-toplevel/>
>>>>    <https://datatracker.ietf.org/doc/draft-ietf-justfont-toplevel/ <https://datatracker.ietf.org/doc/draft-ietf-justfont-toplevel/>>
>>>> 
>>>> 
>>>> 
>>> 
>>>    Of note, the most recent couple of drafts have a new fragment
>>>    syntax for collections (both font/collection and also font/woff2).
>>>    Ken Lunde pointed out that the numeric fragment syntax was
>>>    brittle, as new fonts are typically inserted rather than appended
>>>    into a collection. Instead, a fragment syntax using the PostScript
>>>    name is specified.
> 
> There is a downside to using PS names to identify fonts within a collection, though: it requires the UA to decode/parse all the faces in order to look at their names. The numeric index allows a UA to process only the single face that is actually being requested.
> 
> JK
> 
>>> 
>>>    This syntax was already in use in CSS3 Fonts, for referring to
>>>    locally installed fonts rather than downloaded ones. For use as a
>>>    fragment, the only complication is that six characters are allowed
>>>    in PostScript names and disallowed in fragment identifiers. They
>>>    have to be percent-escaped in the fragments.
>>> 
>>>    An additional benefit is that the syntax is more human readable.
>>>    To get at Foo Bold in a collection (or woff2 of a collection)
>>>    called bar, the syntax is bar.woff2#Foo-Bold for example, not
>>>    bar.woff2#3 or whatever.
>>> 
>>>    As far as I know, no browser or html-to-pdf formatter has support
>>>    for collections. So there is no web compat issue. The new syntax
>>>    will be more usable, and completes what is needed for us to use
>>>    collections in woff2.
>>> 
>>>    --
>>>    Chris