[Fontinst] installing ACaslon - xdvi okay, dvipdf okay, but dvips gives blank output

Lars Hellström Lars.Hellstrom at math.umu.se
Thu Mar 25 15:36:22 CET 2004

At 12.37 +0100 2004-03-21, david.whiting at ncl.ac.uk wrote:
>I am new to this list and new to installing new fonts.  My first
>question is to ask if this is the right place to ask questions about
>problems with creating postscript files using fonts installed via
>fontinst.  If not, does anyone know which list is the correct one?  If
>this is the right list, please read on (I'm afraid this email is a bit

It's a bit off centre (most questions here are rather about running
fontinst), but within the range of topics occuring. Other relevant forums
might be be tex-fonts mailing list (tex-fonts at math.utah.edu) and the
comp.text.tex newsgroup.

>I am using Mandrake Linux 9.2 and I need to use two font families with
>LaTeX and R (an open source implementation of the S statistical
>language).  The fonts are: HelveticaNeue and Adobe Caslon.  I have
>bought the regular fonts, as true type fonts.  HelveticaNeue converts
>okay to postscript with ttf2pt1 and I am able to use the resulting
>postscript files with fontinst.  With ACaslon the resulting files
>cause fontinst to abort with an error message.  I therefore have also
>bought the postscript version of ACaslon-Regular and this works fine
>with fontinst.
>The version of fontinst I am using is:
>Package: fontinst 1998/07/07 v1.801 fontinst installation package
>I have read:
>and am able to view DVI and PDF output, but not postscript.
>Ultimately I also want to be able to use these fonts with R, and at
>the moment I get the following error message from R (this will
>probably not mean much to most people on this list):
>"unknown AFM entity encountered"

Sounds like R encounters something in the AFM file that it doesn't
understand. As I recall it, software reading AFM files is supposed to
simply ignore all lines it doesn't understad (simplifies forward
compatibility), so this looks like a bug in R's AFM parser.

>So here's what I did and found:
>1.  After running fontinst, I followed the steps outlined in the two
>    references mentioned above and can create a dvi file and see
>    text in ACaslon, complete with ligatures.
>2.  dvips shows that the font is being sourced <facr8a.pfb> but when I try
>    to view the postscript file using gv it is blank.

That's odd.

>3.  If I look at the postscript file with vi I can see that
>    ACaslon-Regular is being used (according to %%DocumentFonts) and
>    at the bottom of the file I can see my text (within lots of
>    parentheses):
>    TeXDict begin 1 0 bop 639 566 a Fa(T)-16 b(esting)35
>    b(a)g(ne)r(wl)r(y)g(instal)q(led)g(fo)o(nt)g(in)g(L)r(aT)-16
>    b(eX.)

Yep, that's what the PS produced by dvips looks like. Parentheses are the
PS string delimiters. The words are split up where the distance between
glyphs is not zero (due to e.g. interword space or kerning).

So why doesn't anything show when this postscript is rendered? I'm inclined
to wonder whether there is some PS bug in the font (as .pfb). Have you
tried to produce a postscript file with the ACaslon font using some non-TeX
software? Did it work?

>4.  dvipdf produces a PDF file with ACaslon-Regular, but the ligatures
>    are blank, i.e. missing.  I think this probably be because I do
>    not have the expert font yet.

dvipdf is just dvips+ps2pdf (where ps2pdf, just like gv, is an incarnation
of ghostscript), is it not? Most curious that a conversion to PDF should
have it all working again, but then again, that could happen if the PS font
is broken and the ps->pdf converter is ignoring the incorrect part.

If the ligatures were present when viewing the DVI then they should be
present also in PS and PDF (provided, of course, that the same .pfb or
similar is used in all cases). That the ligatures are missing in the PDF
might alternatively be due to that what you use to view the PDF is buggy
and doesn't like glyphs encoded as e.g. control characters. You might want
to check if the ligatures are the only ones affected, or if also other
glyphs (with similar positions in 8r.enc) are blank.

>13. dvips followed by ps2pdf results in a blank PDF, but if I go to
>    Files->Document Properties->Fonts I see that ACaslon-Regular is
>    the only font listed.

(This would seem to indicate that dvipdf \neq dvips + ps2pdf. But then what
is dvipdf on you system?)

>14. The same thing happens with HelveticaNeue that I converted from
>    TTF to postscript using ttf2pt1.
>I would be grateful for any clues as to how I can create postscript

What happens if you try one of the standard fonts (e.g. Times)? Or PS type
1 CM fonts?

Oh, and then there is always the possibility that you have to turn off font
subsetting in dvips. Not all fonts survive that.

Hope any of the above helps,

Lars Hellström

More information about the fontinst mailing list