[Paul Rubin] > Ehh, I guess I'm not surprised at the slowdown and extra complexity > from the second dict. Oh well. If the module really turns out to be > really used a lot, another (messy) approach would be to write a C > extension that uses a doubly linked list some day.
That seems like an ideal implementation to me. O(1): appending, popping, searching, and deletion O(n): traversal Raymond -- http://mail.python.org/mailman/listinfo/python-list
