problem with epstopdf: shell metacharacters not quoted from $OutputFilename

Hello all,

Apologies if this is the wrong forum for this question; in that case please
point me in the right direction.

System info:

$ uname -a
Linux (somehost.somewhere) 2.6.11-1.14_FC3smp #1 SMP Thu Apr 7 19:38:19 EDT 2005
i686 i686 i386 GNU/Linux
$ rpm -q tetex
tetex-2.0.2-21.3

(Note that while my system has tetex 2.0, I've verified that this problem also
occurs with the epstopdf from the latest tetex-src-3.0.tar.gz).

While using doxygen (v1.4.3) to generate documentation (doxygen calls epstopdf
internally), I came across an apparent problem with epstopdf in which shell
metacharacters (specifically, parentheses) are passed unquoted to a shell pipe
in the following lines of epstopdf:

$ fgrep -1 sOutputFile ~/build/tetex-src-3.0/texk/tetex/epstopdf

  my $pipe = "$GS -q -sDEVICE=pdfwrite $GSOPTS " .
          "-sOutputFile=$OutputFilename - -c quit";
  debug "Ghostscript pipe:", $pipe;
  open(OUT,"|$pipe") or error "Cannot open Ghostscript for piped input";

In my case (using an explicit call to mimic what doxygen is doing internally),
epstopdf is being called as follows:

  epstopdf structfunc__traits_3_01R\(C_1_1_5\)\(\)_4.eps  \
    --outfile="structfunc__traits_3_01R(C_1_1_5)()_4.pdf"

and generates the following message:

sh: -c: line 0: syntax error near unexpected token `('
sh: -c: line 0: `gs -q -sDEVICE=pdfwrite 
-sOutputFile=structfunc__traits_3_01R(C_1_1_5)()_4.pdf - -c quit'

Is this the expected/desired behavior?

>From my admittedly superficial view of the situation, it seems like the
following change would fix the problem

-          "-sOutputFile=$OutputFilename - -c quit";
+          "-sOutputFile='$OutputFilename' - -c quit";

If so, could someone point me to the proper forum for suggesting such a patch?

Or, is it expected that callers of epstopdf should wrap the argument to
--outfile in a second set of quotes, like this:

  epstopdf structfunc__traits_3_01R\(C_1_1_5\)\(\)_4.eps  \
    --outfile="\"structfunc__traits_3_01R(C_1_1_5)()_4.pdf\""

That call works as-is with the current epstopdf, but it feels somewhat
awkward/counter-intuitive to have to wrap the argument in two sets of quotes.

Thanks,
Rob Peters

******************************************************************
Robert J. Peters, Ph.D.              Computer Science/Neuroscience
rjpeters at usc dot edu               Univ. of Southern California
http://ilab.usc.edu/rjpeters/                        3641 Watt Way
tel: 213.740.9223, fax: 626.737.0463         Los Angeles, CA 90089