Graph library for Python

[geremy condra]

On Thu, Dec 10, 2009 at 10:57 AM, Bearophile <bearophileHUGS at lycos.com> wrote:
> Geremy Condra:
>
>> is there a particular way you want your attribution line to read?
>
> You can just use my nickname (in all lowercase), with the list of
> parts you have used. Don't worry.
>
>
>> Well, we all seem to have reinvented the wheel differently ;)
>
> Maybe also because they are designed for different purposes.
>
>
>> Bearophile, Tiago- any interest in trying to combine the
>> best parts of our libraries, with an eye towards eventual
>> integration into the standard library?
>
> The first thing to do is to ask Guido and Hettinger if they are
> willing to put a "good" graph module into the std lib. If their answer
> is positive for some definition of "good", then we can think about
> doing something.
>
> Several years ago I have suggested to put a graph module in the std
> lib, and the answer was something like: "Put the lib online, and if
> people use it a lot, we'll see to put it into the std lib." In the
> meantime my lib was used by no one and ten other graph libs are used
> (networkx seems among the most used), but I think no one of them has
> shown a strong usage. (In the meantime Hettinger has written and added
> two or three or four GOOD data structures to the std lib using a "fast
> lane", avoiding the step of popular usage test).

Well, I've just concluded a short conversation with Raymond Hettinger,
and I think its fair to characterize him as being opposed to the idea
at present. In addition to the popularity test, he's also noted that
ideally a core CPython dev should be involved in the project. Putting
the two together is, AFAICS, a death knell for any extant graph lib.

Having said that, I'd still like to see how much common ground we
could find among the existing libraries. IMHO, there's a lot more
in common than there is different.

Geremy Condra