Hey everyone, I've been trying to come up with an algorithm to describe the assignment of IP subnets. I have something in a proof of concept form that will break a block of addresses into subnets at a user's request. The thing is that the assignments it makes are provably optimal. Within the limits you may place on an assignment, there will never be more than one empty subnet of any particular size. I am curious as to what other things I should try and put into this. Right now, if you request a /16 from a /8, the /16 is considered assigned and further changes within it are not possible. Would 'children' so to speak be useful for you? What would your ideas be? How have you tackled this problem? Feel free to respond off list if you'd like :-)

Appreciate your time!

Mike (sick of spreadsheets) Wiacek

have you looked at/seen RFC 1878? on ftp.isi.edu/pub/bill/tree-2.1.5.tar.gz is a nifty tool. --bill