Thanks, Simon. I hope you find it useful.

The layout algorithm is still a first order hack, and I've discovered at least one tree geometry for which it still needs more work.

I may need to stand back and think more deeply - lots of commercial software provides good and reliable packing of orthogonal trees (with parents centered over child ranges), and I'm sure that there are better algorithms out there if one looked a little harder, or thought a little more carefully.

Who knows - in the fullness of time Omnigroup might even implement one – no word on that yet :-)

(In the meanwhile, my current project involves producing a lot of these trees quite quickly, so I wasn't inclined to wait ...)