[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#156406: dvips executing commands



merge 139085 156406
thanks

"MZ" == Michael Zuther <zuther@min.uni-bremen.de>
"MS" == M. Shell <mshell@ece.gatech.edu>

    MZ> There is a bug in dvips, which prevents for example the
    MZ> generation of an eps-file from a tif-file by the program
    MZ> tiff2ps.

    [deletia]

    MZ> Executing "dvips -o tif-ps-test.ps tif-ps-test.dvi" ends
    MZ> in an error message:

    MZ>   This is dvips(k) 5.86e Copyright 2001 Radical Eye
    MZ> Software (www.radicaleye.com) ' TeX output
    MZ> 2002.08.12:1503' -> bongo.ps <texc.pro><special.pro>. [1
    MZ> dvips.old: Execution of <tiff2ps GRAPHICS/tiger.tif>
    MZ> failed ]


This bug appears to be the same one reported by Barak Perlmutter
in #139085.

A Feburary, 2002, message on the TeX-K list from M. Shell, seems
to sum it up pretty well:

    MS> (1) During the course of validating the new configuration
    MS> file security option, it became apparent that all backtick
    MS> escapes do not work - even in non-secure mode!


    MS> Looking through the code, I traced the problem to line 177
    MS> of output.c:


    MS>  if (secure == 0) {
    MS>     (void)sprintf(errbuf, "Execution of  <%s> failed ", s) ;
    MS>     f = popen(s, FOPEN_RBIN_MODE) ;
    MS>  }
     
     
    MS> Now, in 2000-04-02 the kapthsea library headers were
    MS> changed so that FOPEN_RBIN_MODE is now defined as "rb"
    MS> even on UNIX systems. This is OK with GNU fopen() as it
    MS> will automatically ignore the binary mode directive.
    MS> However, popen() has never tolerated a "b" in the mode
    MS> string (at least on Linux) and will always fail as a
    MS> result.

    MS> So it seems the June 2001 discussion of securing dvips was
    MS> a mute point as it has *really* been secure on UNIX since
    MS> 4/2000 ;)

    MS> The correction is to use a hard-coded "r":

    MS>  f = popen(s, "r");

    MS> and dvips will then work as intended.

The version of dvips that M. Shell is referring to is dvips(k)
5.86f; the dvips in woody is dvips(k) 5.86e.

Provided that this bug has been addressed in later versions of
dvips (and that Thomas Esser includes one of those versions in his
next release), Debian should pick up the fix when we incorporate a
newer teTeX release.

   CMC


[1] <http://www.tug.org/mail-archives/tex-k/2002-February/000419.html>

+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
 Man cannot be civilised, or be kept civilised by what he does in his
	    spare time; only by what he does as his work.
			     W.R. Lethaby
+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
  C.M. Connelly               cmc@debian.org                   SHC, DS
+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+



Reply to: