cross-platform "kern" table (Re: OpenType meeting on September 8th)

[Levantovsky, Vladimir]

Dear all,

With my ISO MPEG AHG hat on, I would like to thank you very much for
your active participation in our meeting yesterday. I believe we had
very productive discussions and accomplished a lot in a limited amount
of time.

I would like to encourage you to continue this email exchange as part of
the ISO ad hoc group activity sending all your messages to the MPEG AHG
email reflector <mpeg-OTspec at yahoogroups.com>. Many of you are already
the members of the MPEG-OTspec group, for those who are not yet
registered on the reflector I would like to invite you to join the group
by accepting the invitation (you will receive an invitation email from
the group) or by visiting the webpage:
http://groups.yahoo.com/group/mpeg-OTspec/join

Thank you again for your active participation.
With kind regards,
Vladimir



> -----Original Message-----
> From: mpsuzuki at hiroshima-u.ac.jp [mailto:mpsuzuki at hiroshima-u.ac.jp] 
> Sent: Tuesday, September 09, 2008 9:38 AM
> To: Julio Gonzalez
> Cc: Michelle Hill; Thomas Phinney; Peter Constable; Mikhail 
> Leonov; Greg Hitchcock; Sergey Malkin; Daniel Fenwick; John 
> Hudson; Levantovsky, Vladimir; Eric Muller; Terence Dowling; 
> Sairus Patel; Simon Daniels; Behdad Esfahbod; David Berlow; 
> Adam Twardoch; Mansour, Kamal; Peter Lofting; John Jenkins
> Subject: cross-platform "kern" table (Re: OpenType meeting on 
> September 8th)
> 
> Dear Sirs,
> 
> In the documentation of new cmap format proposal to compress 
> the cmap in Last Resort font, Apple extension of the "kern" 
> table. The attendances of the meeting agreed that the huge 
> cmap in Last Resort font should be solved and the platform- 
> independent solution is expected, so I think the future 
> direction of new cmap format for the issue would be 
> independent with the platform.
> 
> I was interested in that Apple typography engineers still 
> recognized the Apple extended "kern" table. As I've posted a 
> few posts in OpenType mailing list, it seems that there are a 
> few inconsistencies between the cross-platform format 
> described in OpenType specification and that in Apple's 
> TrueType specification.
> 
> I'm one of the editors of ISO/IEC 9541-4, the standard to 
> interpret typographic properties of ISO OpenType (ISO/IEC
> 14496-22) to exisiting font information interchange framework 
> ISO/IEC 9541, and also the maintainer of Macintosh routines 
> and kern table validator in FreeType2. I want to discuss 
> about how to design a cross-platform kern table for a TrueType font.
> 
> If the description written in OpenType specification is 
> correct, all TrueType (no GPOS but kern table) fonts designed 
> for Microsoft platform should be kerned samely in Apple 
> platform, or, a validator should find the problem.
> I have a few commercial fonts that are kerned on Microsoft 
> platform but not kerned on Apple platform (and FontBook of 
> Mac OS X finds no problem), but yet I could not find the 
> critical difference that causes the difference of kerning.
> 
> I don't have the original TrueType specification by Apple 
> before TrueType GX - I guess, for kern subtable format 0 and 
> 2, there's no difference between old 16bit and new 32bit.
> If my guessing is right, the interpretation of the 16bit 
> coverage element is incompatible between Apple TrueType and 
> OpenType. After a few experiments by mixing kern subtable 
> with Apple-style coverage and Microsoft-like coverage, and I 
> guess current font framework of Mac OS X detects the expected 
> platform per kern table (not per kern subtable):
> 
> * When I put 2 kern format 0 subtables with Apple-style
>   coverage, the kerning values of 2 subtables are summed.
> 
> * When I put 2 kern format 0 subtables with Microsoft-style
>   coverage, the kerning values of 2 subtables are summed.
> 
> * When I put a kern subtable with Apple-style coverage
>   and another kern subtable Microsoft-style coverage,
>   the summation is not executed. Only one subtable is used.
> 
> It is very important to keep the compatibility with existing 
> font resources and font frameworks. As far as no additional 
> functionalities, I don't want to propose new cross-platform 
> kern table (it's already finished in OpenType context - GPOS 
> is that). I want Apple typography engineers to clarify the 
> interpretation algorithm of kern subtables by existing font 
> frameworks, in future update of Apple TrueType specification.
> Is there anybody who can offer consultation?
> 
> Regards,
> suzuki toshiya
> 
> Julio Gonzalez wrote:On Sep 5, 2008, at 9:55 AM, Michelle Hill wrote:
> >We would also like to submit a proposal to extend the cmap table.
> >Please see the enclosed proposal.
>