[XeTeX] Greek Hyphenation (monotoniko)

Jonathan Kew jonathan_kew at sil.org
Wed Jan 11 16:07:10 CET 2006

On 11 Jan 2006, at 2:16 pm, Yves Codet wrote:

> Le 10 janv. 06, à 14:18, Jonathan Kew a écrit :
>> Are you sure it doesn't work? It seems to in my testing.
>> Note that I've just realized that the \showhyphens macro from  
>> Plain TeX isn't useful when working with Unicode fonts in XeTeX;  
>> it won't show you the hyphens that would be found. Could that be  
>> misleading you?
> I have just recompiled formats and tried again: there is no  
> hyphenation after diacritics in the decomposed text, and even in  
> some other cases. Perhaps I made a mistake when I converted it with  
> TECKit. If you have some time for a test I can send you both files  
> (in a private mail, since I suppose the TLG licence does not allow  
> to publicly distribute texts).

What I'm seeing when I try this is that hyphenation after diacritics  
occurs as expected in the decomposed text.... but in the fully  
precomposed text, I am *not* seeing hyphenation after vowels with  
diacritics except in a very few cases.

At first I was expecting identical hyphenation behavior with the  
precomposed and decomposed forms, and was surprised that the  
decomposed version found more hyphens, while the precomposed version  
gave me a bunch of overfull boxes, and only relatively few hyphens.  
But then I realized that some difference should be expected, because  
of the effect of \lefthyphenmin; with decomposed text and  
\lefthyphenmin=2, you can get a hyphen after a single initial vowel  
with diacritic.

Here's a little test using a couple of words from your text:

%%%% start of test file


\setromanfont[Script=Greek]{Galatia SIL}



% force hyphenation
\sloppy \lefthyphenmin=1 \righthyphenmin=1 \hsize=1pt

\hskip0pt νεμοντες % no diacritic

\hskip0pt νέμοντες % precomposed

\hskip0pt νέμοντες % decomposed

\hskip0pt ειδον % no diacritic

\hskip0pt εἶδον % precomposed

\hskip0pt εἶδον % decomposed


%%%% end

A clipping showing the result is attached. Note that with the first  
word, I get a hyphen after the first epsilon in all cases  
(unaccented, precomposed with acute, or with separate acute  
diacritic). But with the second word, the hyphen after iota is  
lacking in the precomposed case.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: texshop_image.pdf
Type: application/pdf
Size: 8995 bytes
Desc: not available
Url : http://tug.org/pipermail/xetex/attachments/20060111/aa472115/texshop_image.pdf
-------------- next part --------------

I'm currently puzzled by this behavior, as I can't see why the  
patterns would fail to work here. So I wonder if there's an  
underlying XeTeX bug that I need to track down.


More information about the XeTeX mailing list