smk (01/16/83)
There has been talk off and on about a program that figures the optimum path to get from machine a to b, using things like path distance, link reliability, link speed, polling times, ... Such a program should have been made an integral part of the news/uucp/USENET software, but since it's not, I wuold like to incorporate into our system. I would greatly appreciate someone sending me the program. However, I think the program should be posted to net.sources. If it is large, it's usefulnes will outweigh the size needed to transmit it. If enough people use it, it will cut down on much of the traffic on the net, by eliminating machines whenever possible and also eliminating the delay.
puder (01/20/83)
I have set up a program (upath-to) on our system that will grep a path from a prepared list of paths, and another program (umail) that preprocesses args that contain '!' and then calls mail. The list of paths is generated by a program (in lisp, of course) that finds a shortest path to a site from a typical uucp map (lines of the form: site<tab>neighbor1 neighbor2 ... ). Note that this is not necessarily the most efficient path. The information needed to find that (such as call frequency, polling direction(s)) is not readily available to me. I would have posted it to the net when I wrote it, but I was expecting it to be obsoleted by the optimal path finder, so I have waited. Has anyone really written an optimal path finder that uses more information than number of sites in the path? Karl Puder burdvax!puder SDC-aBC, R & D Paoli, Pa. (215)648-7555