rnode: strange behavior of relative point translation

Question

I recently discovered a strange behavior of a relative point translation of coordinates given by a node created by \rnode.

The docs tell that the coordinate form ([parameter]node) determines a coordinate relative to the geometrical center of the given node.

I wrote a picture, which output is not consistent with that. At least at first sight :)

Here's the code:

\begin{pspicture}(1,1)

    \rput(0.5,0.5){
      \rnode[t]{n-t}{%
        \rnode[b]{n-b}{%
         \rnode{n-c}{\Huge{X}}%
      }% 
    }%
  }%

  \psdot[dotsize=4pt, linecolor=red](n-t)
  \psdot[dotsize=3pt, linecolor=red](n-c)
  \psdot[dotsize=2pt, linecolor=red](n-b)

  \psdot[dotsize=4pt, linecolor=blue]([offset=0,nodesep=0]n-t)
  \psdot[dotsize=3pt, linecolor=blue]([offset=0,nodesep=0]n-c)
  \psdot[dotsize=2pt, linecolor=blue]([offset=0,nodesep=0]n-b)
\end{pspicture}

that yields this

If I had only read the code it would be clear for me that the blue dots cover the red dots. But not so in the compiled picture.

Is there an error in my reasoning?

Answer 1

that is the internal behaviour of nodes placed by \rnode. Compare the following two nodes:

\documentclass{article}
\usepackage{pst-node}

\begin{document}
\psscalebox{3}{%
\begin{pspicture}[showgrid](4,2)
\psnode(3,1){A}{\Huge A}\psdot([offset=0pt]A) % uses internally \rnode
\pnode(1,1){B}\rput(B){\Huge B}\psdot([offset=0]B)
\end{pspicture}}

\end{document}

However, I'll have a look into the code to see if I can change the behaviour.