Personal info for acli
Name: Ambrose Li
Homepage: http://ada.dhs.org/~acli/
Notes:
2003-07-02:
Is Ghostscript segfaulting in ps2epsi (Bug 686946)? It seems to run fine in XP (comment #1), but I can reproduce it every time on my system, so perhaps my system is hosed, not gs's problem.After I get a SIGSEGV in gdb, if I do up -> break -> c 49 -> watch *((gx_device_memory *) dev).line_ptrs -> c -> break zfillpage -> n -> memory is already corrupted.
I don't know the Ghostscript code well enough to trace it properly, and multi-line macros seem to make the code more difficult to trace.
2003-07-01 (Canada Day):
The .openttcidfont procedure is not present in gs_ttf.ps in 7.33 either. Without downloading more tar files, it would seem that it disappeared from gs_ttf.ps in as of the AFPL 7.x releases (7.20 and above).All the existing docs, as well as automated font installation tools, assume the existence of .openttcidfont; I wonder why the procedure disappeared. Is there a new (undocumented) mechanism to do the same thing? Are the developers unaware of .openttcidfont being important (since it is used by third-party tools)? Or are there GPL-vs-AFPL licensing issues related to this procedure (perhaps with the rest of Ghostscript-CJK)? Without any documentation as to the reason of its disappearance, I don't even know whether it is right to file this as a bug.
KaiSu-Regular (functional again in 8.10) is not functional in 7.33. So I suppose the 7.x AFPL releases should be considered a dead end. However, the fact that normal CJK TrueType fonts have stopped working in the newer AFPL releases still completely baffles me.
Talked to a couple of developers on #ghostscript about the CJK fonts problem, and it seems that loading CJK Truetype fonts indeed are problematic if they are loaded via /Resource. However, it seems that all is well when they are loaded via cidfmap.
Using cidfmap is not obvious after reading Font.html and Use.html; perhaps the docs need some rework.
How do one describe a Unicode font (e.g., Bitstream-Cyberbit) in cidfmap? /CSI seems to limit the font into code pages that can't work across languages. This is problematic, since lots of Japanese fonts can actually be used as Chinese fonts (with some loss of glyphs); also, what to do with Bitstream-Cyberbit, which can support all of C, J, and K?
2003-06-30:
Scratching my head trying to use CJK TTF fonts in Ghostscript 8.10. What used to work in Ghostscript 7.x apparently stopped working as of Ghostscript 8.00, and there seems to be NO DOCUMENTATION as to how to use such fonts in the newer AFPL Ghostscript releases :-(On the brighter side, KaiSu-Regular, which stopped working since Ghostscript 6.x, seems to be working again.
Recent diary entries for acli:
[ Certification disabled because you're not logged in. ]