For tree layout, the way you're plotting it and the layout algorithm go pretty naturally together, but for just the force/spring ones, its not a big deal. I don't really have time or much interest in keeping on working on that package, I probably shouldn't have registered it in the first place. Maybe I should deprecate it.
On Wednesday, February 10, 2016 at 6:13:21 PM UTC-5, hustf wrote: > > There was an inspiring related thread a little while ago: > > > https://groups.google.com/forum/#!searchin/julia-users/graph$20glvisualize/julia-users/ybGrFVKGyDA/KEf6mY0mCwAJ > > My impression is that this subject is full of heuristics, and for graphs > like e.g. the complete Julia package system I believe interactivity is the > most effective heuristics there is. Compose can't offer that so I have > simply used a text file with a blacklist of nodes to remove, and a function > to remove disconnected nodes. > > One of the other heuristics to graph layouts is going 3d, projecting your > nodes on a sphere, a torus etcetera. That's fascinating in itself. You can > start out with a 2d layout and the spring stiffness model or your own > variant of it. But if it never settles into a clarifying layout for the > purpose, you may take that incomplete layout and project it on a spherical > surface and voila you get clarity. Or a torus. This is quite fun. It's > neither lightweight nor easy nor fully compatible with Mac OSX, but a more > complex graph layout application with interactivity could be built using > the current state of GLVisualize. That state, of course, is changing > rapidly. >
