[XeTeX] Japanese Characters in PDF do not match thosein source file.

[Khaled Hosny]

On Tue, Aug 24, 2010 at 04:54:29PM +0900, Andrew A. Adams wrote:
> 
> Andy Lin wrote:
> > Could it be possible you have two copies of IPAMincho in your
> > system/texpath right now? This has been known to result in issues such
> > as incorrect glyphs showing up in the output, as in your case.
> 
> 
> The prize goes to Andy Lin for the simplest solution and the weirdest 
> behaviour of a program I've seen in a while. There were indeed three, not 
> just two, copies of the font on my system. I had one in my local .fonts 
> directory and removed that yesterday, but it didn't seem to help. Today I did 
> a check again and found that two copies were in separate subdirectories in 
> /usr/share/fonts (japanese and IPA). I deleted one of the two and 
> lo-and-behold, I get the attached pdf.
> 
> Thank you all for your suggestions. Please remember this gotcha for anyone 
> having similar problems in future: do a search for the font file and delete 
> duplicates from anywhere XeTeX might find them, particularly from two places 
> in /usr/share/fonts.

That what basically what I said in my reply ealier, but I guess I wasn't
clear enough.

> Seems like an odd error to be caused, though perhaps they were slight 
> variants of the fonts and the xdvipdfmx and acrobat reader were finding the 
> different ones with separate glyph indices or something.

It is a known issue, when you set fonts by font name (as opposed to file
name), XeTeX passes the font name rather than the absolute path to
xdvipdfmx which then have to lookup the font in his own, and if you have
different copies of the font there are chances that xdvipdfmx will find
a different one.

This feels like a FAQ to me, isn't there a place for such FAQs (I don't
think UK-TUG FAQ fits).

Regards,
 Khaled

-- 
 Khaled Hosny
 Arabic localiser and member of Arabeyes.org team
 Free font developer