[tex-k] [ttf2pt1-devel] bug in dvips Re: new ttf2pt1 snapshots

Sergey Babkin babkin@bellatlantic.net
Thu, 22 Nov 2001 23:33:28 -0500

[Resending to tex-k ... --karl]

"Edward G.J. Lee" wrote:
>   BTW, you will need 3.4.0-chinese maps, bsmi00lp55 has some glyphs
>   not in the 3.3.2-chinese maps(such as 110). I think that is why
>   110 is .notdef.

Thanks, this is what I was missing!

>   in bsmi00lp55.t1a here. But this font cannot use by dvips(I'm sorry,
>   the error messages not always the same.). This time I got:
>   <bsmi00lp55.pfb> WARNING: Not all chars found.
>   Debug: Char 0 'd_10470'
>   .....
>   Debug: Char 148 'd_533'
>   The strange thing is, dvipdfm can use the font to profuce .pdf.

OK, I've figured out what is going wrong. Looking at dvips 5.58f,
in the file dvips/t1part.c the function GetToken reads the symbols
from the PostScript file. It contains a bug: it does not consider
'_' (underscore) a valid character to be in a symbol name. (It
does isalpha() and isalnum() and none of these tests returns
TRUE for '_'). You may want to contact the people who work on
dvips and let them know. I've included the address of the author
of t1part.c and of the FreeBSD dvips port maintainer on the CC: list,
so maybe that's enough and they would be able to fix the dvips

In the meantime you can either fix this GetToken() function by yourself
to accept underscores (that's easy) or edit out all the underscores
from the .afm and .t1a files before passing them through t1asm (say,
in vi you can do ":%s/_/Z/g" or you can use sed).


ttf2pt1-devel mailing list