[XeTeX] Strangeness with Minion Pro Opticals Small Caps

Jonathan Kew jonathan_kew at sil.org
Mon Aug 21 11:31:42 CEST 2006


On 21 Aug 2006, at 10:09 am, Ralf Stubner wrote:

> Jonathan Kew <jonathan_kew at sil.org> writes:
> [...]
>> (It may indirectly use the installed version *as well*, because it
>> links dynamically with the installed Fontconfig, and that in turn
>> probably links with Freetype. But this should not affect xetex's
>> rendering.)
>
> ....
>
> However, xdvipdfmx seems to use the system installed version:

True; it has not been "TeX Live-ized" yet.

> Might different freetype versions for XeTeX and xdvipdfmx cause any
> problems?

I very much doubt it, as they don't rely on freetype to interpret OT  
tables or anything like that... only to retrieve the raw data tables  
from the font files. And xdvipdfmx doesn't even have to map  
characters to glyphs; it just draws the glyphs that xetex specifies  
directly by glyph index.

So for Peter to get the "Th" ligature where he should be getting  
"TH" (small-caps), it must be that xetex is applying the ligature  
lookup before the small-caps one. This is done by the ICU library,  
not by Freetype, and *should* behave identically on all platforms,  
given the same .otf font.

It is of course possible that there's a bug in xetex (or the included  
ICU code) that surfaces only on Peter's particular configuration  
(e.g., because of the AMD64?). But the behavior he's seen is  
completely consistent with having incorrectly ordered lookups in the  
GSUB table, which is a plausible error (and would not affect Adobe  
Reader's use of the fonts, so it could easily have gone undetected  
for a while). So I'd like to make very sure whether this is the case,  
before trying to figure out how to track down a possible system- 
dependent bug in the code (without access to a system where it occurs!).

JK



More information about the XeTeX mailing list