W3C home > Mailing lists > Public > www-style@w3.org > September 2010

RE: [css3-fonts] @font-face matching and font-style descriptor

From: Sylvain Galineau <sylvaing@microsoft.com>
Date: Mon, 13 Sep 2010 21:49:33 +0000
To: "www-style@w3.org" <www-style@w3.org>
CC: John Daggett <jdaggett@mozilla.com>
Message-ID: <045A765940533D4CA4933A4A7E32597E27D034EE@TK5EX14MBXC111.redmond.corp.microsoft.com>
(For some reason I can see John's message in the public archive but
it hasn't reached me yet so this may appear in the wrong place).

> Actually, the spec is pretty clear I think:

> (1) local(xxx) references a *single* face with a *fullname* 
>    of "xxx" [Section 4.3]

> (2) style descriptors are used for font matching *not* the underlying
>    style data in the font [Section 4.4]

> IE9 currently looks for a *family* name of "xxx", which is wrong, it
> should be looking for the *fullname*.  The fullname identifies a
> single face uniquely.  Webkit looks up via the fullname but it uses font
> descriptor API's to do the style mapping which results in it basing
> the style mapping on the data in the fonts themselves, which is wrong.

I hadn't considered the name matching issue. That explains it, thanks !

> Probably a few examples of @font-face style matching would probably go
> a long way to clarifying some of this, although only where the spec
> wording makes the matching rules explicit.

Indeed. Also, given its importance in the overall matching algorithm, we
should:
- Consider a more detailed and formal definition of the font name used for
face matching purposes. There is a good background material in Appendix A
but it's informative. I understand some of this relates to OpenType but I
would still rather try to define it so as to depend on it. I do not have
a proposal right now; I frankly am not sure I can come up with one on my
own but I will see what Sergey and I can think of.
- I think the font matching algorithm in 5 should either include this as
well or reference the section that describes the family name matching in
the proper step (#2?).
Received on Monday, 13 September 2010 21:50:08 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:31 GMT