[XeTeX] OpenType Math kerning (cut-ins)

[John Was]

Hello

I guess this is related to the difficulty (impossibility?) mentioned in 
another thread of setting up kerns between different fonts.  I would still 
like to see custom kerns supported within a font, though.  Thus, today I 
noticed that Monotype's open type Baskerville gives collisions at (J, f', 
f), f], etc. (all the usual suspects from the old days).   I'm sure there 
are ways of sorting this out for those who know all about look-up tables and 
suchlike (which I haven't ventured into yet) - but simplest of all would be 
to be able to load a custom kern table that added new kerns to those 
supplied with the font, at the point of loading the font with \font (or one 
of its more sophisticated progeny - but not forgetting the humble \font for 
those of us who stick to plain (Xe)TeX!).

Best


John




----- Original Message ----- 
From: "Jonathan Kew" <jfkthame at googlemail.com>
To: "Unicode-based TeX for Mac OS X and other platforms" <xetex at tug.org>
Sent: Wednesday, October 28, 2009 4:41 PM
Subject: Re: [XeTeX] OpenType Math kerning (cut-ins)


> On 28 Oct 2009, at 09:43, Joel C. Salomon wrote:
>
>> Consider the attached files (requires a recent version of
>> unicode-math).  There does not seem to be any kerning of f_k, even in
>> Cambria Math where it's most obviously needed.
>>
>> I recall from Rhatigan's TUG 2008 presentation "Three typefaces for
>> mathematics" (based on his thesis; see
>> <http://ultrasparky.org/school/pdf/DanielRhatigan_Dissertation.pdf>,
>> pp. 36–37) that OpenType Math fonts have somewhat complicated  kerning
>> (cut-ins) for super-/sub-scripts.  Is support for these implemented in
>> XɘTeX?
>
> Not at the moment. I think this would require some kind of extension  to 
> TeX's model for positioning the super- and subscripts. Currently,  XeTeX 
> just attempts to use the OpenType font within the existing TeX 
> math-layout model, which is somewhat more limited.
>
> The Cambria example does look bad, though; it would be nice to improve 
> this.
>
> JK
>
>