Different spline-styles on different levels not possible?

Is it right, that it is not possible to create edges with different spline-styles in one graph or do I have a mistake somewhere? I'd like to have a graph like the following one with "true splines" from first to second level and straight lines between all other levels. But the line from D1 to D6 is bend, even if I use "splines = false" for this edge.

Digraph G {
node [shape = record];
rankdir = LR;
A0 -> LONGTEXT [splines = true];
A0 -> B1 [splines = true];
A0 -> C1 [splines = true];
A0 -> D1 [splines = true];
A0 -> E1 [splines = true];
A0 -> F1 [splines = true];
A0 -> G1 [splines = true];
A0 -> H1 [splines = true];
A0 -> I1 [splines = true];
A0 -> J1 [splines = true];
B1 -> B2 [splines = false];
C1 -> C2 [splines = false];
C1 -> C3 [splines = false];
C1 -> C4 [splines = false];
D1 -> D2 [splines = false];
D1 -> D3 [splines = false];
D1 -> D4 [splines = false];
D1 -> D5 [splines = false];
D1 -> D6 [splines = false];
H1 -> H2 [splines = false];
I1 -> I2 [splines = false];
J1 -> J2 [splines = false]}

It is true that the splines

It is true that the splines attribute is a graph attribute, and not an edge attribute. However, you can pretty much get what you want by doing

dot in.gv | gvpr -c 'E[$.splines=="false"]{$.pos=""}' | neato -n2 

I didn't use gvpr so far and

I didn't use gvpr so far and I'm not familiar with it's syntax. Trying this advice I get the error message:

gvpr: expected keyword BEGIN/END/N/E...; found ''', line 1

Please tell me what command

Please tell me what command shell environment you are using. Different shells have different quoting conventions.

One way to avoid all of these is to create a file, say spline.gvpr, whose content is

E[$.splines=="false"]{$.pos=""}

and then run dot in.gv | gvpr -c -fspline.gvpr | neato -n2 ...

gvpr in Windows environment

I work with a simple Windows7-"DOS-Box". With the workaroud you suggested I was able to solve my problem with splines and straight lines. But I found an other problem that I will describe in a new thread.

Recent comments