> > I think backward compatibility is a strong argument to keep returning True. >
well, there is also the option to deprecate the is_connected() function for the empty graph and then change the behaviour after a year. By the way, what about defining that the empty graph is connected AND disconnected. ( I'm joking :-) ) But seriously, I just thought, what about dropping the definition for graph connectedness at all. and introduce three different cases instead: - a graph has 0 components - a graph has exactly 1 component - a graph has 2 or more components At least we are able to count ;-) I think If we cannot handle special cases consistently even in theory, we will never ever be able to implement it without getting tons of worms... Jakob Am Donnerstag, 22. August 2013 23:29:45 UTC+2 schrieb Stefan: > > I think backward compatibility is a strong argument to keep returning True. > > I also have an answer based on "my favorite definition is ...", namely the > analogue with matroid connectivity, where any matroid that is too small to > have a k-separation, is automatically k-connected. Extending this to > graphs, the empty graph should be infinitely connected. > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
