Dan, yeah, It i think you are right about the listen coming first. Although it doesn't "really matter" because we've anihalated the runner's threadgroup, and a call to runner.handle would repeatedly generate an exception that would be caught by the catch (IOException) in Webserver's run try block, until listenr = null, which would happen soon afterwards. but obvisouly, it would be nicer to prevent exceptions than just handle them. The only thing I am still not clear on is whether or not the pointers on the Stack threadpool need to be deleted/handled... ashley raiteri
ps. thanks for the all the work you've been doing lately, we might even consider doing a new build and download tar with the frenzy of activity in the last week. -----Original Message----- From: Daniel Rall [mailto:[EMAIL PROTECTED]] Sent: Wednesday, February 20, 2002 1:40 PM To: XML-RPC Development Subject: WebServer shutdown Ashley, shouldn't the listener thread be shut down before the runners so that the server doesn't accept more connections which it can't process? Other than that, this looks good to me. Anyone else have any comments? ---Webserver.java +++Webserver.java @@ -351,7 +351,11 @@ { serverSocket.close(); serverSocket = null; - } + listener = null; + runners = null; + keepalive = false; + } catch (IOException ignore) { } @@ -363,7 +367,21 @@ { - if (listener != null) + if (runners) + { + ThreadGroup R = runners; + runners = null; + try { + R.interrupt(); + } + catch (Exception ex) + { + System.err.println (ex); + ex.printStackTrace(); + } + } + if (listener != null) { Thread l = listener; listener = null; @@ -416,7 +434,7 @@ public void run () { - while (Thread.currentThread() == thread) + while (runners && (Thread.currentThread() == thread)) { con.run (); count++;
