[cups.bugs] [LOW] STR #3641: pstops filter produces no output from gnuplot ps files.

[Carl Michal]

DO NOT REPLY TO THIS MESSAGE.  INSTEAD, POST ANY RESPONSES TO THE LINK BELOW.

[STR Closed w/Resolution]

It is true, if I ask gnuplot to produce postscript (rather than eps), the
resulting file appears to be better behaved.

But shouldn't an eps file print?  Its certainly nice to be able to produce
an eps file to include into other documents and be able to print it alone.

If I create an eps file with inkscape for example, its first line is:
%!PS-Adobe-3.0 EPSF-3.0

and it prints without problem.

ghostscript renders gnuplot's eps files fine, but pstops destroys them. 
If that's due to an error in gnuplot's postscript output then it would be
nice to get it fixed.

For gnuplot, try the following:
> set term post eps  [ or > set term post default ]
> set out 'test.ps'
> plot x**2
> set out

The files you get out of asking gnuplot for postscript vs. eps are very,
very similar.

In fact, changing the first line of the eps file to:
%!PS-Adobe-2.0
does not make the file survive cups' pstops filter.

The key difference between the ps and eps files that gnuplot produces
appears to be the:
%%Page: 1 1
line at the end of the Prolog.

Adding this one line to the eps file makes it work.  Removing it from the
postscript file makes it fail.

I'm no expert on the postscript specification, but I thought eps files had
to be 1 page, and so shouldn't need to have page markers like that.

Link: http://www.cups.org/str.php?L3641
Version: 1.3.11
Fix Version: 1.4.4