结果与并发功能不一致？

Question

I am trying to process lines from a file concurrently, but for some reason I appear to be getting inconsistent results. A simplified version of my code is below:

  var wg sync.WaitGroup
  semaphore := make(chan struct{}, 2)
  lengths:= []int{}

  for _, file := range(args[1:]){
    // Open the file and start reading it
    reader, err := os.Open(file)
    if err != nil {
      fmt.Println("Problem reading input file:", file)
      fmt.Println("Error:", err)
      os.Exit(0)
    }
    scanner := bufio.NewScanner(reader)
    // Start streaming lines
    for scanner.Scan() {
      wg.Add(1)
      text := scanner.Text()
      semaphore <- struct{}{}
      go func(line string) {
          length := getInformation(line)
          lengths = append(lengths, length)
          <-semaphore
          wg.Done()
      }(text)
    }
  }
  wg.Wait()
  sort.Ints(lengths)
  fmt.Println("Lengths:", lengths)

The getInformation function is just returning the length of the line. I then take that line and add it to an array. The issue I'm having is that when I run this multiple times against the same file I get different number of items in my array. I had assumed that since I was using a waitGroup that all lines would be processed every time and therefore the contents of lengths would be the same, but this does not appear to be the case. Can anyone see what I am doing wrong here?

Answer 1

the lengths = append(lengths, length) is getting executed concurrently. This is not safe and will cause problems like missing entries from slice. You can fix this by wrapping the append calls in a mutex, or have the gorountines publish their results to a channel and have a single place that collects them up into a slice.