[XeTeX] Could not load both Bold and BoldItalic.

Peter Dyballa Peter_Dyballa at Web.DE
Thu Dec 4 15:05:34 CET 2008

Am 04.12.2008 um 10:22 schrieb Ulrike Fischer:

>>> OK. Then its "the system" that provides the fonts in a form that  
>>> they
>>> can be used!
>> No – you fooled me! Your example shows the "conventional" side of
>> xdvipdfmx, which comes from dvipdfm (direct convertor from DVI to
>> PDF) and dvipdfmx (which added capabilities to handle complex CJK
>> PostScript fonts) that use TeX fonts, i.e., they work TFM based.
> I don't understand how the tex->xdv-step which needs the TFM can  
> benefit
> from features of xdvipdfmx which is called later.

It can't, I mean: it would be backwards, in time (maybe quantum  
computer will be able to perform this, too; instead of a count  
down ...). XeTeX needs information about the glyphs (characters from  
a particular font) it is going to use. Its primary font interface is  
direct: tables in the font that describe the glyphs in terms of  
dimensions, positions, writing direction, ability to build ligatures,  
etc. OpenType (OT or OTF, based on PostScript or on TrueType),  
TrueType (TT or TTF), Apple Advanced Typography (AAT) fonts provide  
this information. It is used by ICU (International Components for  
Unicode), AAT, or (SIL) Graphite renderers to create shaped text  
(therefore the ability to artificially lengthen/shorten, embold/thin,  
or slant). (Pango, a Greek-Japanese construct for "all tongues," is  
another effort here.) Primitive fonts like PostScript Type 1 and 3  
(?) and 42 do not have any other information than dimensions and an  
arbitrary encoding plus optional creation of ligatures (therefore  
this information can be retrieved as an Adobe Font Metrics, AFM, file).

TeX has its way to create a good layout of a page. Font information  
is passed to it from the fonts used – or from TeX Font Metrics (TFM)  
files if the font is a bare one. With eTeX TeX can "write" in both  
directions. Other extensions are necessary to write up-down. The XDV  
output file is rather primitive (as is DVI, too): use character so- 
and-so from some particular font scaled to some size and put it at  
this position. The output driver uses this information to create some  
other, more in use and also more useful format, because it allows to  
search for text (instead of glyphs or characters).

> Anyway: even if I remove the TFM of utopia (so that pdflatex fails)  
> xelatex still can
> handle the font.

*I* have a very simple explanation based on observation: xdvipdfmx is  
more polite than Japanese! When it fails, it does no harm. I.e., when  
a PDF file existed before it's neither overwritten nor removed by  
xdvipdfmx' failure. Check the PDF file's date and compare it to the  
moment you ran xdvipdfmx!

> Btw: I'm not saying that xetex can handle *every* type1. There are
> certainly some which are so old, buggy, incomplete or special that  
> xetex
> will fail.

Then these fonts aren't PostScript Type 1.

Mit friedvollen Grüßen


Banken sprengen heißt Sonne rein lassen.

More information about the XeTeX mailing list