On Monday, July 9, 2012 2:59:33 AM UTC+1, Birk Eisermann wrote: > > (I have searched and found that there has been discussion on how much > networkx will be adapted in sage - and it seems that there is some issue > with the license... right? I also read the sage-devel thread "graph > theory: refactor implementation of spanning tree algorithms" which give > me some impression...) >
Huh? Sage <= 5.1 includes NetworkX 1.2, which is released under BSD [1], and there is a much needed upgrade to NetworkX 1.6 coming in time for Sage 5.2; we have been shipping an outdated version of NetworkX because of API changes that broke Sage tests, not due to license issues [2]. >From Sage, we can use networkx as one possible choice for graphs backends, and even when it is not chosen there are many graph methods (clustering, centrality, topological ordering...) that do nothing but call the corresponding NetworkX function. The main "problem" with NetworkX is that it is a pure python system focusing on having easily readable and usable code, and this has some performance issues, particularly for large networks. There have been a few discussions about using Cython to speedup code but so far nobody has bit the bullet. Much of Nathan's work in the graphs classes had to do with implementing fast algorithms in Cython. I have no strong opinion on how to refactor the humungous graph classes, but perhaps we could benefit from pushing some of our cython code upstream to NetworkX rather than having it in the sage library. Cheers, J [1] http://networkx.lanl.gov/reference/legal.html [2] http://trac.sagemath.org/sage_trac/ticket/12806 -- -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
