Just when you think you have it all covered. I was testing the solution to the serial port issue I wrote about a couple of days ago.

When I ran the code, the CPU pegged at 100%. When I went back to the code it hit me like the proverbial metric ton of bricks. Have a look at this routine:

private string PeekResponseLine()

{
    while (true)
    {
        string bufferCopy = string.Empty;
        lock (incomingBuffer)
        {
            bufferCopy = incomingBuffer;
        }

        int length = bufferCopy.IndexOf("\r\n");
        if (length > -1)
        {
            if (bufferCopy.Contains(GsmResponses.MessageTextWanted))
            {
                length = GsmResponses.MessageTextWanted.Length;
            }
            else if (bufferCopy.Contains(GsmResponses.MessageReceived))
            {
                length = bufferCopy.IndexOf("\r\n", GsmResponses.MessageReceived.Length);
                if (length > -1)
                    length += 2;
            }
            else
                length += 2;

            if (length > -1)
            {
                string response = bufferCopy.Substring(0, length);
                return response;
            }
        }
    }

}


What do you notice based on what I've told you about the CPU utilisation? It struck me pretty much straight away. The while(true) {} loop will just sit there iterating, waiting for a CRLF to come. This is terribly shoddy work on my part. Of course I know you shouldn't poll. I left the loop there, but added an AutoResetEvent that is triggered off data arriving from the serial port. That way, the loop only iterates when there's a reason to.

The result? CPU utilisation went from 100% to 1%. It's not every day you get a perf increase like that. Even if it is your own stupid fault...