Build and run it:

go build tsunami server tsunami client tsunami help

This is a golang implementation of tsunami-udp:

http://tsunami-udp.sourceforge.net/

I found a good c-based implementation, but it didn't look like a good fit for simply adding go bindings:

https://github.com/sebsto/tsunami-udp

Here's a good description with some psuedo-code:

http://tsunami-udp.cvs.sourceforge.net/viewvc/tsunami-udp/docs/howTsunamiWorks.txt?revision=1.5&view=markup

Here's the psuedo-code from that page:

     **Server**
 start
 while(running) {
   wait(new incoming client TCP connection)
   fork server process:
   [
     check_authenticate(MD5, "kitten");
     exchange settings and values with client;
     while(live) {
       wait(request, nonblocking)
       switch(request) {
          case no request received yet: { send next block in sequence; }
          case request_stop:            { close file, clean up; exit; }
          case request_retransmit:      { send requested blocks; }
       }
       sleep(throttling)
     }
   ]
 }


**Client**
 start, show command line
 while(running) {
    read user command;
    switch(command) {
       case command_exit:    { clean up; exit; }
       case command_set:     { edit the specified parameter; }
       case command_connect: { TCP connect to server; auth; protocol version compare;
                               send some parameters; }
       case command_get && connected:  { 
           send get-file request containing all transfer parameters;
           read server response - filesize, block count;
           initialize bit array of received blocks, allocate retransmit list;
           start separate disk I/O thread;
           while (not received all blocks yet) {
              receive_UDP();
              if timeout { send retransmit request(); }

              if block not marked as received yet in the bit array {
                 pass block to I/O thread for later writing to disk;
                 if block nr > expected block { add intermediate blocks to retransmit list; }
              }

              if it is time { 
                 process retransmit list, send assembled request_retransmit to server;
                 send updated statistics to server, print to screen;
              } 
           }
           send request_stop;
           sync with disk I/O, finalize, clean up;
       }
       case command_help:    { display available commands etc; }
    }
 }

Some testing commands:

time seq -f "%g" 1 20000 | parallel echo {} > /dev/udp/localhost/1200 for qqq in seq 1 20; do echo a $qqq ; done | parallel --pipe --line-buffered -N 1 -L 0 --round nc -u localhost 1200 for qqq in seq 1 20; do echo a $qqq ; done | parallel --pipe cat - > /dev/udp/localhost/1200