![](https://secure.gravatar.com/avatar/86776c6c595af5117de5ba7b41bc33b5.jpg?s=120&d=mm&r=g)
Morten Kjeldgaard <mok@imsb.au.dk>:
simple dict-based stuff -- more performance is needed). kjbuckets looks like a nice alternative, as does the Boost Graph Library (not
Kjbuckets is *very* nice indeed.
Yes, I guess it is. But the project doesn't seem very active...
It is a compact and very fast implementation. I don't see why you'd want to wrap this functionality into NumPy, which has a very well-defined scope and an efficient implentation. It would be a shame to bloat it with something which is discretely different.
Yes, I guess you're right. There is no point in adding this sort of thing to numarray. My motivation for using numarray in my implementations was simply that it would mean that the necessery tools would be (or might be in the future ;) available in the standard distribution.
I have modified kjbuckets so that it compiles and works with Python 2.x. You can pick it up at
ftp://xray.imsb.au.dk /pub/birdwash/packages/Python2.1/SRPMS/python-kjbuckets-2.2-7.src.rpm
Just do "rpm --rebuild" on it.
I sent the patch to the original author, but it appears he is no longer maintaining it. Never mind, it works great.
Well... I do sort of mind... I'm a bit wary of using unmaintained software. Not that I would never do it or anything... But I think it would be a bonus to use stuff that is being actively maintained and developed. But I guess I'll take another look at it. (Any idea where the "kj" prefix comes from, by the way?)
/Morten
-- Magnus Lie Hetland The Anygui Project http://hetland.org http://anygui.org