summaryrefslogtreecommitdiff
path: root/docs/tutorials/013/page01.html
diff options
context:
space:
mode:
Diffstat (limited to 'docs/tutorials/013/page01.html')
-rw-r--r--docs/tutorials/013/page01.html65
1 files changed, 0 insertions, 65 deletions
diff --git a/docs/tutorials/013/page01.html b/docs/tutorials/013/page01.html
deleted file mode 100644
index 2f5c50afaf9..00000000000
--- a/docs/tutorials/013/page01.html
+++ /dev/null
@@ -1,65 +0,0 @@
-<!-- $Id$ -->
-<HTML>
-<HEAD>
- <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
- <META NAME="Author" CONTENT="James CE Johnson">
- <TITLE>ACE Tutorial 013</TITLE>
-</HEAD>
-<BODY TEXT="#000000" BGCOLOR="#FFFFFF" LINK="#000FFF" VLINK="#FF0F0F">
-
-<CENTER><B><FONT SIZE=+2>ACE Tutorial 013</FONT></B></CENTER>
-
-<CENTER><B><FONT SIZE=+2>Multiple thread pools</FONT></B></CENTER>
-<P>
-<HR WIDTH="100%">
-<P>
-My intent with this tutorial was to derive from ACE_Data_Block instead
-of ACE_Message_Block so that we could leverage the reference counting
-nature of that object.
-<P>
-Along the way, I sort of got distracted... What I ended up with is a
-poor excuse for ACE_Stream that implements a simple state machine.
-<P>
-The application is built around a thread pool where the pool's svc()
-method takes work units from the message queue for processing. As
-each unit is taken from the queue, the process() method is invoked to
-do some work. The twist is that after processing the message, we
-enqueue it into another thread pool to do more work. This continues
-through a chain of thread pools until the last where the unit's fini()
-method is called for finishing up any outstanding work.
-<P>
-The chain of thread pools is uni-directional using a singly-linked
-list of Task derivatives. Each pool has the same number of tasks in
-order to keep things simple.
-<P>
-Kirthika's abstract:
-<UL>
-In this tutorial, a singly linked list of thread-pools, each of which is
-a subtask and which acts as the finite state machine node, is used to
-simulate a finite state machine.
-<P>
-A task is created with a number of subtasks. Once the message block is
-obtained from the queue, it is verified to see whether a task has a
-subtask. If so, it is forwarded to the subtask. Thus the mesage
-traverses over the whole list. As a safety measure for destroying the
-block after it goes through the whole list, an effective and simple
-Memory Leak Detector has been implemented. It is a counter which
-increments when the object where it resides is created and decrements on
-its deletion.
-<P>
-Another optimisation from the previous tutorials on Message Queues, is
-the bundling of the Data block within the Message Block. The Data block
-provides reference counting, so duplication of data is avoided. It is
-deleted only when its reference count drops to zero. Now updating
-this count between threads call for synchronisation and in comes the
-ACE_Mutex, a lock which takes care that the counting is thread-safe.
-<P>
-Although the example isn't a full-fledged Finite State Machine,
-i.e. it has to be tweaked to be able to jump states on different inputs,
-it definitely proves to be a great lesson and introduces us to quite a
-few new ACE classes and the ways they can be mixed and matched to
-produce the end-system desired.
-</ul>
-<P><HR WIDTH="100%">
-<CENTER>[<A HREF="../online-tutorials.html">Tutorial Index</A>] [<A HREF="page02.html">Continue This Tutorial</A>]</CENTER>
-