preferred udp ports?


    Jun 26 2006 | 6:11 am
    Maybe its been asked...
    I've been using udpsend quite a lot lately with a rewire-enabled collective i've built to send a variety of stuff to a few other computers. I've noticed quite often I get "udp packet dropped" errors in the max window of the collective which is sending outwards. I have two ports open, the rewire interface is on 7000 and some midi data from the sequencer is on 7001. Both ports use osc tags. port 7000 is essentially banging hostsync~ every 1ms, so its sending quite a lot of data, without a tap.allchange or what not for lists being that the remote applications may need to be opened and closed. I guess I want to see an end to the packet drops. Are there any ports that are recommended or does it not matter?
    cheerio,
    binez0r

    • Jun 26 2006 | 9:43 am
      On 26-Jun-2006, at 8:11, binez0r wrote:
      > I guess I want to see an end to the packet drops. Are there any
      > ports that are recommended or does it not matter?
      I don't think the ports make a difference. The network, other network
      traffic, amount of bandwidth you want, things like that are important.
      Don't forget that UDP is a no-guarantees protocol. If a packet
      doesn't get through, UDP basically shrugs its shoulders and says
      "tough luck." If you want guarantees, you either gotta build 'em
      yourself on top of UDP or use TCP or something.
      -- P.
      -------------- http://www.bek.no/~pcastine/Litter/ -------------
      Peter Castine +---> Litter Power & Litter Bundle for Jitter
      Heavy-Duty Mathematics for Everyday Use
      iCE: Sequencing, Recording &
      Interface Building for |home | chez nous|
      Max/MSP Extremely cool |bei uns | i nostri|
    • Jun 27 2006 | 5:14 am
      About the only recommendation on ports is not to use ports that the
      system already claims--which is why 7000 is a safe choice. The first
      256 are definitely reserved, and you will get an error if you try to
      claim a port that is already in use. But otherwise, there are no speed
      or reliability gains (though personally I would favor prime numbers
      and Fibonacci numbers, esp. Fib primes, and eschew powers of two
      altogether, no telling what havoc they could wreak).
      -- Paul
      --
      ----- |(*,+,#,=)(#,=,*,+)(=,#,+,*)(+,*,=,#)| -----