# [tex-k] bug in either fourier-bb or in dvips

Reinhard Kotucha reinhard.kotucha at web.de
Thu Jul 13 01:47:37 CEST 2006

>>>>> "Karl" == Karl Berry <karl at freefriends.org> writes:

>     now this resulting eps fil has the wrong bounding box (it's
> way too wide).  Indeed, I get %%BoundingBox: -20 709 100 718 which
> is 120bp wide.  Kind of a big character :).

>     I do not know whether this is caused by fourier-bb or dvips.

> Well, adding \setbox0=\hbox{N}\showbox0 shows that the tfm width
> is 7.54999pt, which seems fine.

It's absolutely correct.  The line

53 755 hsbw

in the disassembled font file denotes that the width is 755pt at a
fontsize of 1000pt, hence 7.55pt at 10pt.

> (Not proof of anything, I know, but indicative.)

It proves that the tfm file is correct.

> I'm not completely sure if there is something in the character
> description which goes way out of bounds, but I don't think so.
> I'll attach the output I got from t1disasm fourier-bb.pfb.

I extracted the font from the eps file generated by dvips and
calculated the bounding box of the glyph, which is 0 0 703 693.  This
sounds reasonable and I get same results when I load the font into
fontforge.  I also do not see anything abnormal in the disassembled
font.  t1lint only complains that some hints (BlueValues) are missing.

Testing dvips with other fonts and testing older versions of dvips
could be helpful to isolate the problem.

Regards,
Reinhard

--
----------------------------------------------------------------------------
Reinhard Kotucha			              Phone: +49-511-4592165
Marschnerstr. 25
D-30167 Hannover	                      mailto:reinhard.kotucha at web.de
----------------------------------------------------------------------------
Microsoft isn't the answer. Microsoft is the question, and the answer is NO.
----------------------------------------------------------------------------