[DVIPDFMx] type1 subsetting bug

[Jin-Hwan Cho]

Hi, Karl. It was a good source for me to tackle the well-known bug in  
Mac OS X.

However, (x)dvipdfmx uses CFF format for type1 to reduce the size of  
PDF.
So "lenIV" variable can be found only in the code reading type1 fonts,  
not writing
the fonts. Therefore, I still wonder whether the attached patch solves  
the bug or not,
because the PDF result made by dvipdfmx has the same bug.

Anyway I will check again the CFF writing codes.

Best regards, ChoF.

On Mar 24, 2009, at 9:11 AM, Karl Berry wrote:

> Jonathan, ChoF, et al. --
>
> A bug in subsetting Type 1 fonts was found a few days ago.  It's been
> around for years.  See
> http://tug.org/pipermail/pdftex/2009-March/007975.html.
>
> I don't see the expression "t1_lenIV > 0" anywhere other than pdftex,
> luatex, dvips, and metapost (it's been fixed in all of those), but I
> thought I should forward it anyway just in case it might be relevant  
> to
> xetex or (x)dvipdfm(x) nonetheless.
>
> Thanks,
> Karl
>
> --- writet1.c   (revision 562)
> +++ writet1.c   (working copy)
> @@ -1492,7 +1492,7 @@
>         cr = 4330;
>         cs_len = 0;
>         return_cs = xtalloc(t1_lenIV + 1, byte);
> -        if (t1_lenIV > 0) {
> +        if (t1_lenIV >= 0) {
>             for (cs_len = 0, r = return_cs; cs_len < t1_lenIV; cs_len 
> ++, r++)
>                 *r = cencrypt(0x00, &cr);
>             *r = cencrypt(CS_RETURN, &cr);
>
> _______________________________________________
> dvipdfmx mailing list
> dvipdfmx at ktug.or.kr
> http://project.ktug.or.kr/mailman/listinfo/dvipdfmx