billhathaway

Thanks to Tyler Burnell, Bill Kennedy and others who had written all the boilerplate I copied in main.go and main_test.go

Note: the benchmark data is horribly unrealistic (10M input of either no matches or all matches), but I think the model of testing MB/sec throughput of large input sets is the right one for the problem statement.  I welcome additions where we make the test data more realistic.  Using a very small data set of a few hundred bytes doesn't seem like a good test
for stream processing algorithm performance.

go test -run none -bench . -benchtime 3s -benchmem 
testing: warning: no tests to run
PASS
BenchmarkProcessByteUnmatched-8      	      30	 143059391 ns/op	  69.90 MB/s	       1 B/op	       1 allocs/op
BenchmarkProcessByteMatched-8        	      30	 120619527 ns/op	  82.91 MB/s	       1 B/op	       1 allocs/op

billhathaway

 package main

 import "fmt"
  
   var (
     version  = "develop"
     otherval = "unset"
   )

billhathaway

package psutil

import (
	"fmt"
	"log"
	"strings"

	"cloud.google.com/go/pubsub"
	netcontext "golang.org/x/net/context"
)