[tex-k] Re: Encoding clashes with dvips(k) 5.90a

[Michail Vidiassov]

Dear All,

Tomas G. Rokicki wrote:

> Encoding problem with dvips fixed?
> The problem was, a font was being included in an EPSF file, yet
> dvips was not including the whole font in the output file as it
> should (since it has no way of knowing precisely what characters
> the EPSF file used).  This will also happen with old versions of
> dvips if you use characters in the EPSF file that are not also
> used in the TeX file.

> I've got a patch, but I would want to make sure the patch is
> tested fairly well before releasing a dvips that includes it.

Unfortunately, not for me.

A LaTeX file has $-1$ and includes a metapost output file,
with a label $-1$ in it. The first minus sign is displayed,
but not the second one (from included file).
The problem seems to be caused by the fact, that the minus sign
is double encoded, dvips uses code \241 and downloads the necessary
part of the font, but metapost uses code \000.

It seems to fall under the case of "characters in the EPSF file
that are not also used in the TeX file".

The fix mentioned in the referenced post did not help.

A test case with dvi and ps files is at
http://www.iaas.msu.ru/tmp/bugtest.tar


                       Sincerely, Michail
The source files follow.
-------------------------------
bugtest.tex:

\documentclass[a4paper,dvips]{article}

\usepackage{epsfig}

\begin{document}
\psfull
\thispagestyle{empty}
$-1$

\epsfig{file=bugtest.1}

\end{document}
-----------------------------
bugtest.mp:

beginfig(1);

label(btex $-1$ etex , (40,40) );
endfig;

end