This surprises me, since I've mentioned it in the README file of the CJK package, already in the second paragraph.
I apologize again that I seldom join mailing lists, and the CJK package were included in the distribution so I didn't even know the README exist. There're just too many READMEs littering around on a Linux box, you know.
Please check the pdftex mailing list archive and try to download the latest version from Sarovar -- probably compiling with the `-g -O0' compiler flags to activate debugging, and to disable optimization. In case this doesn't fix the issue, submit a bug report to the pdftex maintainers.
Will try your suggestion at a later time. Thank you.
Looking into arialuni version 0.84 and cyberbit 1.1 (this is what I have on my box) I can see that the latter has a real glyph for uniFF0C, while the former uses the `comma' glyph as a subglyph. Maybe this doesn't work properly with pdftex.
How can I check it myself? So that if later on I get a strange font I can verify myself for potential problems. Mere visual inspection doesn't tell, you know.
As a workaround I suggest that you use the subfonts.pe script from CJK 4.6.0 (or the latest snapshot) to split arialuni.ttf in a set of Type 1 subfonts. Be warned that it takes hours even on a fast computer until the conversion has finished! The proper SFD file is Unicode.sfd; Some time ago I've posted an updated version to this list; you might find it also in the `freetype1-contrib' CVS repository (available from savannah.gnu.org/projects/freetype).
Thanks. In fact your updated Unicode.sfd that support the full Unicode range brought me to this mailing list in the first place. The subfonts script completed in 3 minutes instead of hours on my system. Is it normal? But I copied the PFB/ENC/TFM to the texmf tree and regenerating the PDF. U+FF0C seems to appear now, and the segfault disappeared (so I didn't bother to upgrade pdftex at this point anyway if it's still usable). That's marvellous. Will verify again on the other system I have at a later time.
Thank you so much again for your input. Regards, Bernard Chan. _______________________________________________ Cjk maillist - [email protected] http://lists.ffii.org/mailman/listinfo/cjk
