[XeTeX] Typographical question: xetex' default wordspacing (\fontdimen2--7)

Ralf Stubner ralf.stubner at physik.uni-erlangen.de
Sun Jun 25 12:17:27 CEST 2006

"Toralf Senger" <Senger at mpimp-golm.mpg.de> writes:
>>| \comment{The remaining code in this section sets various metric 
>>| parameters for the font. Ideally, all these parameters should already  
>>| have been set; the values computed here are merely crude guesses 
>>| about what might be the right value.}
> Interesting! So who sets those parameters IDEALLY? And based on what
> reasoning? 

Ideally the font designer would specify them, just as he specifies the
width of space. However, I don't know of any possibility to specify
these parameters in a font (short of using the nonstandard 'ftpm' sub-
table of the also nonstandard 'TeX' table introduced by FontForge
<URL:http://fontforge.sf.net/non-standard.html#TeX>). So normally the
person creating the metrics for TeX using fontinst should specify them. 

> That may be true. But I think fontdimens are not the right place to
> worry too much about the paragraph-setting capabilities of TeX. There
> are amble of other parameters which controll that: 
> \penalty, \linepenalty, \hyphenpenalty, \exhyphenpenalty,
> \adjdemerits, \doublehyphendemerits, \finalhyphendemerits,
> \pretolerance, \tolerance, \emergencystretch,\looseness

Sure, there are many other possibilities to influence TeX's paragraph
breaking, but the fontdimens belong into this list, too. IMO having
reasonable defaults for these values is important.

> If one does not want to do
> that and a paragraph is really bad with lines sticking into the
> margings, on could put a \sloppy in front of that (which to make TeX
> relax a bit (and make spacing worse)... 

Due to missfeature (one could call it a bug ;-) in TEX's paragraph
breaking algorithem, \sloppy can produce worse results than \fuzzy on
some paragraphs. :-(

>>1/2 is definitely appropriate for most English texts.
> For German as well.

I used German only as an example were I happen to know the problems. I
am more interested in other scripts, which might have very different


More information about the XeTeX mailing list