git.squeep.com Git - reservoir_sample/blob - test/validate-statistics.sh

   1 #!/bin/bash
   2 #
   3
   4 # Collects and counts the output of multiple invocations over the same input stream.
   5 # Counts ought to be roughly the same, based on quality of random source.
   6
   7 invoke_n() {
   8         seq $1 | ./reservoir_sample -r /dev/random 2>/dev/null
   9 }
  10
  11 invoke_n_i() {
  12         local i=0
  13         while [ $i -lt $2 ]
  14         do
  15                 invoke_n $1
  16                 ((i++))
  17                 local note=`expr $i % 1000`
  18                 if [ $note -eq 0 ]
  19                         then
  20                         echo "... iteration $i of $2" 1>&2
  21                 fi
  22         done
  23 }
  24
  25 tally_n_i() {
  26         invoke_n_i $1 $2 | sort -n | uniq -c
  27 }
  28
  29 tally_n_i 20 10000