On Sun, Apr 26, 2009 at 11:03 PM, Søren Hauberg <so...@hauberg.org> wrote: > søn, 26 04 2009 kl. 22:03 +0200, skrev Radek Salač: >> I'm realy sorry i sent you a wrong archive last time. >> I'm sending you new archive, it's completly new files no modification >> in current code. > > No problem. I don't have time right now to play with your code, but I > quick look told me that things looked nice and clean :-) However, you > seem to be missing license information from your code.
It was only working version, I§m sending the new one with licence info. >> Well, I also don't like dependencies but I was thinking about it quite long. >> Octave now as far as I know is using GNU plot. It's a really strong >> tool but for directed graphs we need other approach. >> Now if I have Adjacency matrix I must find the best location for every >> single node, and edge. It's quite hard because there must be some >> minimum distance between edges and nodes, We want to have "pretty" >> graph, with minimum of crosing edges. We can't just split nodes on >> screen and draw lines between them. This computation is really >> difficult. > > Yeah, I realise this is non-trivial. For me, the problem with using > Graphviz is not the extra dependency, but the lack of integration with > the rest of Octave. It would be really nice if we use the standard > plotting tools as this would allow graphs to appear as subplots, having > their layout changed using the standard get/set method, etc. > I understand and i make litle refactoring now the name is dotplot because it just generate description in dot language. Which can by used by graphviz or latex. > About node-placement then I guess we could use multidimensional scaling > which is really trivial to implement in Octave. This should provide a > decent (but not perfect) placement. It would, however, require an > implementation of the all-pairs shortest path algorithm. I think that this should be in another procedure. Node placement is only par of the problem, there is lot of missing feature for this task in gnuplot. Some are trivial (drawing node as a elipse :-) ). Some will be more complicated. Arrow between nodes which shouldn't cros if posible. But I will look at it in future. > That being said, since your the one doing the programming you should > make the decisions :-) > >> Well this is completely different from normal plots. Maybe a >> completely new package will be the best location for it if you accept >> my code in octave forge. > > We could have a 'graph' package with graph related algorithms (shortest > path, minimum spanning tree's, etc). Your stuff would fit well into such > a package. Some time ago there was talk about creating such a package, > but I don't remember the details right now. > > Søren > > Radek Salac
dotplot.tar.gz
Description: GNU Zip compressed data
------------------------------------------------------------------------------ Register Now & Save for Velocity, the Web Performance & Operations Conference from O'Reilly Media. Velocity features a full day of expert-led, hands-on workshops and two days of sessions from industry leaders in dedicated Performance & Operations tracks. Use code vel09scf and Save an extra 15% before 5/3. http://p.sf.net/sfu/velocityconf
_______________________________________________ Octave-dev mailing list Octave-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/octave-dev
