Jin-Hwan Cho wrote: > On Sep 1, 2005, at 12:13 AM, Ralf Stubner wrote: > > If you have the source code of DVIPDFM, check the file "pdfobj.c". > In the line 631, the function pdf_check_name() checks the validity > of the character "#". However, the variable "valid_chars" does not > contain "#".
Thanks a lot! The rather trivial patch is attached and seems to compile well with the sources from teTeX 3. Could somebody with the file in question test if this is indeed all that's needed? >>> Of course. Can dvipdfmx be used as a drop-in-replacement for >>> dvipdfm, >>> even if you don't process CJK files? Are there any disadvantages? >> >> If that where possible, it would be ideal, since dvipdfm seems to be >> pretty dead upstream. My personal experiences with dvipdfmx for non >> CJK >> texts has been pretty good, but I haven't made any extensive tests. > > For more than 3 years, DVIPDFMx have been extensively tested by many > users. There might be some bugs which are not found yet. But the authors > of DVIPDFMx try to fix immediately if reported to them. Sounds good. Maybe we should contact Thomas Esser and suggest switching to dvipdfmx? cheerio ralf
--- pdfobj.c.orig 2005-08-31 18:47:17.999074000 +0200 +++ pdfobj.c 2005-08-31 18:46:30.779074000 +0200 @@ -631,7 +631,7 @@ int pdf_check_name(const char *name) { static char *valid_chars = - "!\"$&'*+,-.0123456789:;[EMAIL PROTECTED]|~"; + "!\"#$&'*+,-.0123456789:;[EMAIL PROTECTED]|~"; if (strspn (name, valid_chars) == strlen (name)) return 1; else