When we compile this code:


    % Some variants of closing the path
    \coordinate (a) at (0,0) ;
    \coordinate (b) at (2,0) ; 
    \draw[] (a) .. controls +(0,1) and +(0,1) .. (b) -- (a);

    \coordinate (c) at (3,0) ;
    \coordinate (d) at (5,0) ; 
    \draw[] (c) .. controls +(0,1) and +(0,1) .. (d) -- cycle;

    \coordinate (e) at (6,0) ;
    \coordinate (f) at (8,0) ; 
    \draw[] (e) .. controls +(0,1) and +(0,1) .. (f) -- (e) -- cycle;

    % Without closing the path
    \coordinate (g) at (0,-1) ;
    \coordinate (h) at (2,-1) ; 
    \draw[] (g) .. controls +(0,1) and +(0,1) .. (h);

    \coordinate (i) at (3,-1) ;
    \coordinate (j) at (5,-1) ; 
    \draw[] (i) .. controls +(0,1) and +(0,1) .. (j) 
            (j) -- (i);

This is the output we expect and get using pdflatex (TeXLive 2014, TikZ 3.0.0):

enter image description here

When we print the PDF, however, we get this:

enter image description here

The same happens when converting the PDF with pdftops (that is how I obtained the image).

We would certainly expect Bézier curves to print in all cases. What happens here and how do we work around or fix this?

  • I've not tried to print it, but compiled with pdflatex (MikTeX) and converted to ps with pdf2ps, the result is correct. – Ignasi Dec 11 '14 at 17:52
  • Looks like a driver or PDF viewer problem though. That's very interesting to see. – percusse Dec 11 '14 at 18:05
  • 3
    Using TeXLive 2014 I can replicate the problem when viewing a postscript file generated using pdftops with both evince and gv. However, pdf2ps produces a postscript file which renders correctly in both viewers. – Mark Wibrow Dec 11 '14 at 21:38
  • 1
    I get the same results as @MarkWibrow except that my viewer is Okular. pdftops is provided by poppler (at least on my system) while pdf2ps is part of ghostscript. Something to do with the cairo dependency? – cfr Dec 11 '14 at 22:56
  • 1

Following the comments above, I have confirmed that pdf2ps produces correct PS files. Therefore, pdftops seems to have a bug, which cfr has reported.

  • I'm on poppler/pdftops 0.24.5 while the latest stable release is 0.31.0. The bug cfr reported has not been marked as fixed, but I can not say for sure that it persists in newer versions. – Raphael Feb 11 '15 at 13:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.