On Mon, 17 Jul 2006 05:53:04 -0400, Marcus Watts wrote: >Randomness leading up to >> > There are no useful answers for idiots. >> >> I like that phrase, I'll have to remember that one. > >Just for the hell of it, I'll try offering a few >useless answers. > > { it's clear the originator isn't worried about secrecy > or anonymity, given he's using a remote radius server > and asked for help in a public forum. } > > { if he *was* interested in privacy & anonymity, surely he'd > be exploring broadcast or unidirectional protocols such as > digital radio mondiale and not asking us questions. } > >1. I'm pretty sure Vincent Cerf didn't intend for any tcp protocols to >survive changing the IP address every minute. Although a lot of his >work seems to have involved machines that were too heavy to carry and >too expensive to re-address every minute, he appears to have >nevertheless been keenly interested in mobile computing & radio use >before either were common. I've no doubt he'd be amused by the >originator's attempt, though I doubt he'd be supportive. The problem >does sound remarkably like a "worst case" roaming scenario with >wireless IP. Maybe something involving a revolving restaurant? > > { Since the originator of this thread appears to have been > relying on what are presumably non-dedicated data circuits & > shared servers, his connections are subject to random delay > depending on competition from other user(s) of those services. > Excessive delay will surely lead to lost data, and snippets > that cannot be pasted together without weirdness. > Presumably those delays will get worse with time... } > >2. If you *were* trying to piece together a reliable data feed >out of very short snippets, you'd probably have much better luck >if you managed up to *two* separate overlapping connections -- >dropping one once you've sync'd up with the other. Dropping >duplicated data is easier than recreating lost data. > >3. If you wanted to use internet protocols to give you a reliable >feed (instead of making one yourself as in 2), you'll want to run >a vpn on top of your physical connection, so that you can then >use tcp to manage packet drops due to the underlying connection >randomly disappearing. > >4. "sox" will concatenate mp3 input's together. You'd then need to >re-encode the output stream using some mp3 encoder. sox won't >be capable of recovering data lost due to network drops, >and it's not going to help you with pasting snippets together either. >There is tons of other audio software that can do the same thing, >with variable amounts of fluff and bother. > >5. There are a bunch of people who are very keen on matching audio >fragments up. Some phrases they like to use are "audio finger-printing", >or "automatic music identification". Unfortunately these are also the >very same people who tend to be real keen on proprietary data & >software techniques. Fortunately for you, the patent process is >"supposed" to encourage people to provide sufficient information to >make it possible to make experimental use of patented technology. >Unfortunately for you, "supposed to" to a lawyer is rather like what >"possible" means to a mathematician who is asked if the product >of large primes can be factored. > > -Marcus Watts
What a beautiful piece of writing. There are chunks that I cannot claim expertise on. Even they sound plausible (in the non-derogatory sense) and the bits that I do know about seem consistant with reality. Marcus, it was a joy to read a well constructed essay with no ad hominem bits that should, but I would not bet my lefty on it, be the end of this tiresome thread. Or at least the end of the discursive part, you may see other compliments. ;-) >From the land "down under": Australia. Do we look <umop apisdn> from up over? Do NOT CC me - I am subscribed to the list. Replies to the sender address will fail except from the list-server. Your IP address will also be greytrapped for 24 hours after any attempt. I am continually amazed by the people who run OpenBSD who don't take this advice. I always expected a smarter class. I guess not.