From: Robert Klemme <shortcutter@...>
Date: 2011-10-28T18:04:09+09:00
Subject: Re: How to write TCPServer code that can exit?

On Fri, Oct 28, 2011 at 10:11 AM, Chris Lervag <chris.lervag@gmail.com> wrote:
> I've got a bit of a problem with my network code. Perhaps someone can
> help me out with this:
>
> I've got a program in which I run a TCP server. The code is basically
> like this:
>
> def start_server
>  s = TCPServer.new(@port)
>  loop do
>    Thread.start(s.accept) do |session|
>      # handle incoming connections...
>    end
>  end
> end
>
> The server code works just fine. However, my problem is that as it is,
> Im having trouble testing it. Once the server is initiated, I can't shut
> it down in code. Thus Im not able to figure out how to do rspec examples
> with it. If I try to start the server code in an rspec example and run
> some client stuff to test it, I never get to the next examples, because
> the server code wont quit. It just runs forever.
>
> Is there a better way to do this, so I can shut it down at will, and be
> able to make a proper test suite?

I think you need to generally think about how that server will be shut
down.  There are a few options I can think of off the top of my head:

 - if the protocol allows it reserve a message for shutdown so clients
can terminate the server
 - write a signal handler which cleanly shuts down the server (for
example close listening socket, wait for all client threads to
terminate); to support that your server could write out its pid to a
well known location (e.g. /var/run).

Kind regards

robert

-- 
remember.guy do |as, often| as.you_can - without end
http://blog.rubybestpractices.com/