@c ======================================================================== @c Other Network Topologies @c ======================================================================== @node Other-network-topologies @chapter Other Network Topologies @cindex topology @cindex Channel @cindex NetDevice @cindex topology!bus @cindex topology!point-to-point @cindex PointToPointChannel @cindex PointToPointNetDevice @emph{Network topology} is the study of the arrangement of of the elements (in @command{ns-3} represented by the classes @code{Channel} and @code{Node}) of a network. Two fundamental types of physical topologies are the @emph{point-to-point} and @emph{bus} topologies. We have already been exposed to the @command{ns-3} channel specialization named @code{CsmaChannel}. This is a simulation of a bus network. We also provide a simulation of a point-to-point channel with associated net devices. As described previously, the associated C++ classes specialize the @command{ns-3} base classes @code{NetDevice} and @code{Channel} and are called @code{PointToPointNetDevice} and @code{PointToPointChannel} respectively. We will use combinations of these bus and point-to-point topology elements to show how to create several commonly seen network topologies. @section A Point-to-Point Network We're going to take what might be seen as a step backward and look at a simple point-to-point network. We will be building the simplest network you can imagine. A serial link (point to point) between two computers. When you see this point-to-point network, you can think of an RS-422 (or RS-232 for you old-timers) cable. This topology is shown below. @sp 1 @center @image{pp,,,,png} @cindex CreateObject @cindex InternetNode We have provided a file for you in the @code{tutorial} directory called @code{tutorial-point-to-point.cc}. You should now be familiar enough with the system to pick out fairly easily what has been changed. Let's focus on the following lines: @verbatim Ptr n0 = CreateObject (); Ptr n1 = CreateObject (); Ptr link = PointToPointTopology::AddPointToPointLink ( n0, n1, DataRate (38400), MilliSeconds (20)); PointToPointTopology::AddIpv4Addresses (link, n0, "10.1.1.1", n1, "10.1.1.2"); @end verbatim You can see that we created two @code{InternetNode} objects in the usual way. Then, instead of creating a @code{CsmaChannel} we create a @code{PointToPointChannel}. This point-to-point channel, which we call @code{link}, connects node zero (@code{n0}) and node one (@code{n1}) over a simulated link that runs at 38400 bits per second and has a 20 millisecond simulated speed-of-light delay. This call also creates appropriate net devices and attaches them to nodes zero and one. We then add IP addresses to the net devices we just created using the topology helper @code{AddIpv4Addresses}. Node zero gets the IP address 10.1.1.1 and node one gets the IP address 10.1.1.2 assigned. The alert tutorial user may wonder what the network number or prefix is of those IP addresses. The point-to-point topology assumes that you want a @code{/30} subnet and assigns an appropriate net mask for you. It then then @emph{asserts} that the network numbers of the two net devices match. So there is an implicit network mask created down in the topology code that looks like, @verbatim Ipv4Mask netmask("255.255.255.252"); @end verbatim The rest of the code you should recognize and understand. We are just going to echo one packet across the point-to-point link. You should be now be able to build and run this example and to locate and interpret the ASCII trace file. This is left as an exercise for you. The file @code{tutorial-point-to-point.cc} is reproduced here for your convenience: @verbatim /* -*- Mode:C++; c-file-style:"gnu"; indent-tabs-mode:nil; -*- */ /* * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 as * published by the Free Software Foundation; * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include "ns3/log.h" #include "ns3/ptr.h" #include "ns3/internet-node.h" #include "ns3/point-to-point-channel.h" #include "ns3/mac48-address.h" #include "ns3/point-to-point-net-device.h" #include "ns3/point-to-point-topology.h" #include "ns3/udp-echo-client.h" #include "ns3/udp-echo-server.h" #include "ns3/simulator.h" #include "ns3/nstime.h" #include "ns3/ascii-trace.h" #include "ns3/pcap-trace.h" #include "ns3/global-route-manager.h" NS_LOG_COMPONENT_DEFINE ("PointToPointSimulation"); using namespace ns3; // Network topology // // point to point // +--------------+ // | | // n0 n1 // int main (int argc, char *argv[]) { LogComponentEnable ("PointToPointSimulation", LOG_LEVEL_INFO); NS_LOG_INFO ("Point to Point Topology Simulation"); Ptr n0 = CreateObject (); Ptr n1 = CreateObject (); Ptr link = PointToPointTopology::AddPointToPointLink ( n0, n1, DataRate (38400), MilliSeconds (20)); PointToPointTopology::AddIpv4Addresses (link, n0, "10.1.1.1", n1, "10.1.1.2"); uint16_t port = 7; Ptr client = CreateObject (n0, "10.1.1.2", port, 1, Seconds(1.), 1024); Ptr server = CreateObject (n1, port); server->Start(Seconds(1.)); client->Start(Seconds(2.)); server->Stop (Seconds(10.)); client->Stop (Seconds(10.)); AsciiTrace asciitrace ("tutorial.tr"); asciitrace.TraceAllQueues (); asciitrace.TraceAllNetDeviceRx (); Simulator::Run (); Simulator::Destroy (); } @end verbatim @section A Star Network A point-to-point network is considered a special case of a star network. As you might expect, the process of constructing a star network is an extension of the very simple process used for a point-to-point link. We have provided a file for you in the @code{tutorial} directory called @code{tutorial-star.cc} that implements a simple star network as seen below. @sp 1 @center @image{star,,,,png} In order to create a star network, we need to be able to instantiate some number (greater than one) of net devices on a node. In the name of simplicity of use, the @code{PointToPointTopology} topology helper does not allow one to do this. We provided a separate topology helper class, the @code{PointToPointIpv4Topology} helper class that provides the slightly finer granularity we need to accomplish a star network. In order to use this new helper we have to load the definitions by including the appropriate file. @verbatim #include "ns3/point-to-point-ipv4-topology.h" @end verbatim The star that we're going to create has a node in the center (@code{n0}) with six nodes surrounding (@code{n1} - @code{n6}). You should be able to easily find and understand the code that creates these nodes. @verbatim Ptr n0 = CreateObject (); Ptr n1 = CreateObject (); Ptr n2 = CreateObject (); Ptr n3 = CreateObject (); Ptr n4 = CreateObject (); Ptr n5 = CreateObject (); Ptr n6 = CreateObject (); @end verbatim Next, we get into the differences between the @code{PointToPointTopology} helper and the @code{PointToPointIpv4Topology} helper. The @code{PointToPointIpv4Topology} helper looks and feels a little like the @code{CsmaIpv4Topology} helper. Just like you created a CSMA channel previously, you need to create a point-to-point channel. The following code creates a @code{PointToPointChannel} and calls it @code{link01}. You can interpret this name as being the channel (or @emph{link}) from node zero to node one. @verbatim Ptr link01 = PointToPointIpv4Topology::CreateChannel (DataRate (38400), MilliSeconds (20)); @end verbatim You need to provide a data rate for the channel which we set at 38400 bits per second. You must also provide a speed-of-light delay which we set at 20 milliseconds. Just as you added a net device to the nodes in the CSMA tutorial section, you do the same here but with a point-to-point net device. The following code illustrates how we do that: @verbatim uint32_t nd01 = PointToPointIpv4Topology::AddNetDevice (n0, link01); @end verbatim We call the @code{PointToPointIpv4Topology} helper and ask it to add a net device to node zero (@code{n0}) and connect it to the appropriate point-to-point link (@code{link01}) which you will recall is the serial link from node zero to node one. If you look at the following code, you will see the same calls are repeated to create the remaining five point-to-point channels and connect them to net devices on node zero. The next new code is found after the ``spokes'' of the star have been created. It looks like the following: @verbatim uint32_t nd1 = PointToPointIpv4Topology::AddNetDevice (n1, link01); uint32_t nd2 = PointToPointIpv4Topology::AddNetDevice (n2, link02); uint32_t nd3 = PointToPointIpv4Topology::AddNetDevice (n3, link03); uint32_t nd4 = PointToPointIpv4Topology::AddNetDevice (n4, link04); uint32_t nd5 = PointToPointIpv4Topology::AddNetDevice (n5, link05); uint32_t nd6 = PointToPointIpv4Topology::AddNetDevice (n6, link06); @end verbatim Here we are creating the net devices on the nodes surrounding the center node. In the first call, we are adding a net device on node one (@code{n1}) and connecting that net device to the channel named @code{link01}. Remember that we created the channel @code{link01} as the channel connecting node zero and node one. We previously created a net device on node zero and attached that device to @code{link01}. Here we are connecting the other side of that link to node one. The return value from this call is the net device index of the created net device. The next section of code adds addresses to the net devices we just created. The first call adds the IP address 10.1.1.1 to the net device going from node zero to node one. Recall that we first created a node named @code{n0} and a channel called @code{link01}. We added a net device to @code{n0} and remembered the net device index as the @code{uint32_t nd01}. This meant the net device @emph{nd} on node @emph{0} that we connected to node @emph{1}. We call @code{AddAddress} to add an IP address (10.1.1.1) to the net device on node zero identified by the net device index @code{nd01}. We provide a net mask suitable for a point to point network. This is typically a /30 address but we don't force that in this API. After setting up the address on node zero, we do the same for the node on the other end of the ``spoke'' --- in this case node one, with its single net device. Note that the network number is the same on both sides of this network. @verbatim PointToPointIpv4Topology::AddAddress (n0, nd01, "10.1.1.1", ``255.255.255.252''); PointToPointIpv4Topology::AddAddress (n1, nd1, "10.1.1.2", ``255.255.255.252''); @end verbatim The following code repeats this pattern assining similar IP addresses to the remaining net devices. Note that there are no @code{Mac48Address} address assignments --- they are not required. The rest of the code you should recognize and understand. We are just going to echo one packet across the point-to-point link. You should be now be able to build and run this example and to locate and interpret the ASCII trace file. This is left as an exercise for you. The file @code{tutorial-star.cc} is reproduced here for your convenience: @verbatim /* -*- Mode:C++; c-file-style:"gnu"; indent-tabs-mode:nil; -*- */ /* * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 as * published by the Free Software Foundation; * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include "ns3/log.h" #include "ns3/ptr.h" #include "ns3/internet-node.h" #include "ns3/point-to-point-channel.h" #include "ns3/mac48-address.h" #include "ns3/point-to-point-net-device.h" #include "ns3/point-to-point-ipv4-topology.h" #include "ns3/udp-echo-client.h" #include "ns3/udp-echo-server.h" #include "ns3/simulator.h" #include "ns3/nstime.h" #include "ns3/ascii-trace.h" #include "ns3/pcap-trace.h" #include "ns3/global-route-manager.h" NS_LOG_COMPONENT_DEFINE ("StarSimulation"); using namespace ns3; // Network topology // // n3 n2 // | / // | / // n4 --- n0 --- n1 // / | // / | // n5 n6 int main (int argc, char *argv[]) { LogComponentEnable ("StarSimulation", LOG_LEVEL_INFO); NS_LOG_INFO ("Star Topology Simulation"); Ptr n0 = CreateObject (); Ptr n1 = CreateObject (); Ptr n2 = CreateObject (); Ptr n3 = CreateObject (); Ptr n4 = CreateObject (); Ptr n5 = CreateObject (); Ptr n6 = CreateObject (); Ptr link01 = PointToPointIpv4Topology::CreateChannel (DataRate (38400), MilliSeconds (20)); uint32_t nd01 = PointToPointIpv4Topology::AddNetDevice (n0, link01); Ptr link02 = PointToPointIpv4Topology::CreateChannel (DataRate (38400), MilliSeconds (20)); uint32_t nd02 = PointToPointIpv4Topology::AddNetDevice (n0, link02); Ptr link03 = PointToPointIpv4Topology::CreateChannel (DataRate (38400), MilliSeconds (20)); uint32_t nd03 = PointToPointIpv4Topology::AddNetDevice (n0, link03); Ptr link04 = PointToPointIpv4Topology::CreateChannel (DataRate (38400), MilliSeconds (20)); uint32_t nd04 = PointToPointIpv4Topology::AddNetDevice (n0, link04); Ptr link05 = PointToPointIpv4Topology::CreateChannel (DataRate (38400), MilliSeconds (20)); uint32_t nd05 = PointToPointIpv4Topology::AddNetDevice (n0, link05); Ptr link06 = PointToPointIpv4Topology::CreateChannel (DataRate (38400), MilliSeconds (20)); uint32_t nd06 = PointToPointIpv4Topology::AddNetDevice (n0, link06); uint32_t nd1 = PointToPointIpv4Topology::AddNetDevice (n1, link01); uint32_t nd2 = PointToPointIpv4Topology::AddNetDevice (n2, link02); uint32_t nd3 = PointToPointIpv4Topology::AddNetDevice (n3, link03); uint32_t nd4 = PointToPointIpv4Topology::AddNetDevice (n4, link04); uint32_t nd5 = PointToPointIpv4Topology::AddNetDevice (n5, link05); uint32_t nd6 = PointToPointIpv4Topology::AddNetDevice (n6, link06); PointToPointIpv4Topology::AddAddress (n0, nd01, "10.1.1.1", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n1, nd1, "10.1.1.2", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n0, nd02, "10.1.2.1", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n2, nd2, "10.1.2.2", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n0, nd03, "10.1.3.1", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n3, nd3, "10.1.2.2", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n0, nd04, "10.1.4.1", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n4, nd4, "10.1.4.2", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n0, nd05, "10.1.5.1", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n5, nd5, "10.1.5.2", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n0, nd06, "10.1.6.1", "255.255.255.252"); PointToPointIpv4Topology::AddAddress (n6, nd6, "10.1.6.2", "255.255.255.252"); uint16_t port = 7; Ptr client = CreateObject (n0, "10.1.1.2", port, 1, Seconds(1.), 1024); Ptr server = CreateObject (n1, port); server->Start(Seconds(1.)); client->Start(Seconds(2.)); server->Stop (Seconds(10.)); client->Stop (Seconds(10.)); AsciiTrace asciitrace ("tutorial.tr"); asciitrace.TraceAllQueues (); asciitrace.TraceAllNetDeviceRx (); Simulator::Run (); Simulator::Destroy (); } @end verbatim @subsection Routing If you are really excited about this simulator you may have already tried to modify the scripts outside the tutorial. I know that one of the first things that would have occurred to me when I saw the star network would have been to start trying to add applications to echo packets from nodes other than zero. If you tried, for example, to start the echo client on node one instead of node zero, you would have found an empty trace file. The reason for this is that you have now created an internetwork. This means you will need to enable internetwork routing. We have provided a file for you in the @code{tutorial} directory called @code{tutorial-star-routing.cc} to show you how this is done. This extremely tricky and difficult change is shown below: @verbatim GlobalRouteManager::PopulateRoutingTables (); @end verbatim This one-line addition, located just before the simulation runs, tells the @command{ns-3} @emph{global route manager} to walk the topology you created and build internetwork routing tables for all of the nodes in the simulation. We changed the client application so that it runs on node four: @verbatim Ptr client = CreateObject (n4, "10.1.1.2", port, 1, Seconds(1.), 1024); @end verbatim Now if you build and run @code{tutorial-star-routing.cc} you can examine the @code{tutorial.tr} file and see that your UDP echo packets are now correctly routed through the topology. @section A Dumbbell Network One of the most interesting simple topologies (from a phenomenological point of view) is commonly called a dumbbell network. The name derives from a superficial similarity in form to a piece of exercise equipment. The dumbbell model is typically composed of two bus or star network elements connected via a point-to-point link. The point-to-point link is usually configured with a lower bandwidth than the bus elements to provide a @emph{choke point}. The following is a representation of the topology. @sp 1 @center @image{dumbbell,,,,png} We have provided a file that constructs this dumbbell network and creates enough data flowing across the choke point that some packets will be dropped. The file is called @code{tutorial-linear-dumbbell.cc} and is located in the @code{tutorial} directory. We have already covered all of the code used to create this network, so we will just quickly go over the main sections of the script. The first section creates a CSMA lan that will become the left side of the dumbbell network. This code should be very familiar since we used the same process to create our first example. @verbatim // // Create the lan on the left side of the dumbbell. // Ptr n0 = CreateObject (); Ptr n1 = CreateObject (); Ptr n2 = CreateObject (); Ptr n3 = CreateObject (); Ptr lan1 = CsmaTopology::CreateCsmaChannel (DataRate (10000000), MilliSeconds (2)); uint32_t nd0 = CsmaIpv4Topology::AddIpv4CsmaNetDevice (n0, lan1, "08:00:2e:00:00:00"); uint32_t nd1 = CsmaIpv4Topology::AddIpv4CsmaNetDevice (n1, lan1, "08:00:2e:00:00:01"); uint32_t nd2 = CsmaIpv4Topology::AddIpv4CsmaNetDevice (n2, lan1, "08:00:2e:00:00:02"); uint32_t nd3 = CsmaIpv4Topology::AddIpv4CsmaNetDevice (n3, lan1, "08:00:2e:00:00:03"); CsmaIpv4Topology::AddIpv4Address (n0, nd0, "10.1.1.1", "255.255.255.0"); CsmaIpv4Topology::AddIpv4Address (n1, nd1, "10.1.1.2", "255.255.255.0"); CsmaIpv4Topology::AddIpv4Address (n2, nd2, "10.1.1.3", "255.255.255.0"); CsmaIpv4Topology::AddIpv4Address (n3, nd3, "10.1.1.4", "255.255.255.0"); @end verbatim The code to generate the CSMA lan on the right side is similar; only the names have been changed. @verbatim // // Create the lan on the right side of the dumbbell. // Ptr n4 = CreateObject (); Ptr n5 = CreateObject (); Ptr n6 = CreateObject (); Ptr n7 = CreateObject (); Ptr lan2 = CsmaTopology::CreateCsmaChannel (DataRate (10000000), MilliSeconds (2)); uint32_t nd4 = CsmaIpv4Topology::AddIpv4CsmaNetDevice (n4, lan2, "08:00:2e:00:00:04"); uint32_t nd5 = CsmaIpv4Topology::AddIpv4CsmaNetDevice (n5, lan2, "08:00:2e:00:00:05"); uint32_t nd6 = CsmaIpv4Topology::AddIpv4CsmaNetDevice (n6, lan2, "08:00:2e:00:00:06"); uint32_t nd7 = CsmaIpv4Topology::AddIpv4CsmaNetDevice (n7, lan2, "08:00:2e:00:00:07"); CsmaIpv4Topology::AddIpv4Address (n4, nd4, "10.1.2.1", "255.255.255.0"); CsmaIpv4Topology::AddIpv4Address (n5, nd5, "10.1.2.2", "255.255.255.0"); CsmaIpv4Topology::AddIpv4Address (n6, nd6, "10.1.2.3", "255.255.255.0"); CsmaIpv4Topology::AddIpv4Address (n7, nd7, "10.1.2.4", "255.255.255.0"); @end verbatim Next, we create a point to point link to connect the two lans. We connect the point-to-point channel between nodes three (on the left lan) and four (on the right lan). You should recoginze this as substantially similar to the link setup from the @code{point-to-point} example. @verbatim // // Create the point-to-point link to connect the two lans. // Ptr link = PointToPointTopology::AddPointToPointLink ( n3, n4, DataRate (38400), MilliSeconds (20)); PointToPointTopology::AddIpv4Addresses (link, n3, "10.1.3.1", n4, "10.1.3.2"); @end verbatim Then we configure data flows. We create four echo clients that send UDP packets from the left side lan to servers created on the right side lan. Notice that we send 100 packets with an inter-packet gap of ten milliseconds instead of the single packet we have previously used. This data rate is sufficient to saturate the point-to-point link and will cause packets to be dropped when the queue on the link net devices overflows (the default maximum queue depth is 100 packets). Note that we stagger the start of the echo clients to slowly bring up the data rates. @verbatim // // Create data flows across the link: // n0 ==> n4 ==> n0 // n1 ==> n5 ==> n1 // n2 ==> n6 ==> n2 // n3 ==> n7 ==> n3 // uint16_t port = 7; Ptr client0 = CreateObject (n0, "10.1.2.1", port, 100, Seconds(.01), 1024); Ptr client1 = CreateObject (n1, "10.1.2.2", port, 100, Seconds(.01), 1024); Ptr client2 = CreateObject (n2, "10.1.2.3", port, 100, Seconds(.01), 1024); Ptr client3 = CreateObject (n3, "10.1.2.4", port, 100, Seconds(.01), 1024); Ptr server4 = CreateObject (n4, port); Ptr server5 = CreateObject (n5, port); Ptr server6 = CreateObject (n6, port); Ptr server7 = CreateObject (n7, port); server4->Start(Seconds(1.)); server5->Start(Seconds(1.)); server6->Start(Seconds(1.)); server7->Start(Seconds(1.)); client0->Start(Seconds(2.)); client1->Start(Seconds(2.1)); client2->Start(Seconds(2.2)); client3->Start(Seconds(2.3)); server4->Stop (Seconds(10.)); server5->Stop (Seconds(10.)); server6->Stop (Seconds(10.)); server7->Stop (Seconds(10.)); client0->Stop (Seconds(10.)); client1->Stop (Seconds(10.)); client2->Stop (Seconds(10.)); client3->Stop (Seconds(10.)); @end verbatim The remainder of the file should be quite familiar to you. Go ahead and run @code{tutorial-linear-dumbbell}. Now take a look at the trace (@code{tutorial.tr}) file. You will now see trace lines that begin with @code{d}. Alternatively you can search for the string ``queue-drop'' which is the expansion of the drop code ('d'). Interpretation of a dropped packet is straightforward. We have expanded the first @code{queue-drop} trace for you below. See the section on ASCII tracing for details. @verbatim 00 d 01 2.40938 02 nodeid=3 03 device=1 04 queue-drop 05 pkt-uid=124 06 LLCSNAP(type 0x800) 07 IPV4( 08 tos 0x0 09 ttl 63 10 id 20 11 offset 0 12 flags [none] 13 length: 1052) 10.1.1.3 > 10.1.2.3 14 UDP(length: 1032) 15 49153 > 7 16 DATA (length 1024) @end verbatim We leave it as an exercise to examine the trace files in more detail. @c ======================================================================== @c Nonlinear Thinking @c ======================================================================== @node Nonlinear-Thinking @chapter Nonlinear Thinking One thing that all of our examples so far have in common is that they are composed of a linear collection of calls into the @command{ns-3} system. The programmers among the readers may have wondered why there is not as much as a for-loop in all of the examples. The answer is that we wanted to introduce you to @command{ns-3} scripting with a minimum of conceptual overhead. We're going to remedy that situation shortly. We have written a number of @command{ns-3} scripts in C++. Although we have been perfectly linear in our script implementations, just like any other C++ program, an @command{ns-3} script can use any features of the language you desire. If you will look back at the @code{tutorial-linear-dumbbell.cc} example, you may notice that the code to create the left and right sides of the dumbbell is operationally identical --- only the names change. An obvious improvement of this program would be to use subroutines to create the sides. Since we are working with C++, we should probably do this in an object-oriented way. Since object-oriented design is somewhat of a black art to some people, we'll take some time here and outline a simple methodology you can follow. @section Object Design 101 --- Class Ipv4BusNetwork If you are a master of object oriented design, feel free to skip or skim this section, in which we derive a simplistic but fully operational bus network class. So you want to create a BusNetwork class. Often the biggest hurdle in a design is figuring out how to get started. One of the simplest and most straightforward ways to do an object decomposition of a problem is to simply write down a description of the problem and take a look at the words you used. Let's take some time and do that, first at a very high level. @example A bus network is an implementation of a particular network topology that contains some number of nodes. Each of these nodes is attached to a single multi-drop channel. The network itself has some attributes independent of the topology such as a network mask, network number (prefix) and base IP address. @end example The first thing to do is to focus on the nouns and adjectives. These will give you a starting point for required classes and member variables. Immediately we can notice that at the highest level we are talking about the noun @emph{network}. This probably won't surprise you. We also have an adjective that modifies the noun --- @emph{bus}. This should lead us to our first class defintion. Usually class names are constructed in the same way as an English language sentence would be spoken. For example, one would speak of a @emph{bus network} in conversation, so we would normally create a @code{class BusNetwork} to represent it. One thing to note is that we have used two words in our description quite naturally: @emph{is} and @emph{has}. When you see these words should should immediately think of the object-oriented concepts of @emph{ISA} (inheritance) and @emph{HASA} (containment) respectively. We wrote that a bus network @emph{is} an implementation of a particular network topology. Perhaps you will agree that there is a natural base class called @code{Network} that @emph{has} the attributes discussed above. The fact that a @code{BusNetwork} @emph{ISA} kind of @code{Network} suggests inheritance. Let's capture that thought right away remembering that we're focused on IP version four here: @verbatim class Ipv4Network { public: Ipv4Address m_network; Ipv4Mask m_mask; Ipv4Address m_baseAddress; }; class Ipv4BusNetwork : public Ipv4Network { }; @end verbatim Let's take a look at the @emph{HASA} relationships of the bus network. Clearly it will @emph{have} a reference to the underlying channel that implements the actual communications medium. We use smart pointers for those references, so one member variable is obvious: @verbatim Ptr m_channel; @end verbatim A bus network will also need to contain references to all of the nodes we eventually want to create. If you are working in C++ and see the words contain or container, you should immediately think of the Standard Template Library or STL. A quick search of the available containers there will probably lead you to consider the vector class. A vector is a container that looks like an array. This is just what we need here. Again, we want to use smart pointers to reference our nodes, so the declaration of the vector would look like, @verbatim std::vector > m_nodes; @end verbatim It will save you headaches in the future if you notice that the space between the two right brackets is required to differentiate this situation from a right-shift operator. So we have a pretty good start already after just a little work. Now we need to turn our attention to actions. Let's write another little description of the things you consider doing to a Bus network. @example We need to be able to create a bus network. We need to be able to delete a bus network. We need to be able to get a handle to a node in order to add applications. We need to be able to set the network, mask and base address somehow, specify how many nodes to create and provide the underlying channel its required bandwidth and delay parameters. @end example We now look at the @emph{verbs} in that sentence. These will give a good starting point for the methods of the classes. For example, the verbs @emph{create} and @emph{delete} should suggest @emph{constructor} and @emph{destructor}. The verb @emph{get} leads us to providing a method called @code{GetNode}. We have to provide a number of parameters so we can either provide @emph{setters} or we can simply pass them in as parameters to our constructors. Since this is a simple example, we won't bother to implement getters and setters (methods to get and set member variables to enhance data hiding). Let's use this guidance to finish up our class declarations: @verbatim class Ipv4Network { public: Ipv4Network (Ipv4Address network, Ipv4Mask mask, Ipv4Address address); virtual ~Ipv4Network (); Ipv4Address m_network; Ipv4Mask m_mask; Ipv4Address m_baseAddress; }; class Ipv4BusNetwork : public Ipv4Network { public: Ipv4BusNetwork ( Ipv4Address network, Ipv4Mask mask, Ipv4Address startAddress, DataRate bps, Time delay, uint32_t n); virtual ~Ipv4BusNetwork (); Ptr GetNode (uint32_t n); private: std::vector > m_nodes; Ptr m_channel; }; @end verbatim That's it. We have actually already walked through almost all of the code required to construct a bus network in our @code{tutorial-csma-echo.cc} example, so let's just jump forward and take a look at an implementation of this thing. We provide an implementation for you in the files @code{ipv4-bus-network.h} and @code{ipv4-bus-network.cc} located in the @code{tutorial} directory. We also provide an example that uses the new class in the file @code{tutorial-bus-network.cc}. The interesting method from our current perspective is the Ipv4BusNetwork constructor, shown below: @verbatim Ipv4BusNetwork::Ipv4BusNetwork ( Ipv4Address network, Ipv4Mask mask, Ipv4Address baseAddress, DataRate bps, Time delay, uint32_t n) : Ipv4Network (network, mask, baseAddress) { Ipv4AddressGenerator::SeedNetwork (mask, network); Ipv4AddressGenerator::SeedAddress (mask, baseAddress); m_channel = CsmaTopology::CreateCsmaChannel (bps, delay); for (uint32_t i = 0; i < n; ++i) { Ptr node = CreateObject (); uint32_t nd = CsmaIpv4Topology::AddIpv4CsmaNetDevice (node, m_channel, Mac48Address::Allocate ()); Ipv4Address address = Ipv4AddressGenerator::AllocateAddress (mask, network); CsmaIpv4Topology::AddIpv4Address (node, nd, address, mask); m_nodes.push_back (node); } } @end verbatim Notice that we do the simple and straightforward thing and pass all of our parameters to the constructor. For those unfamiliar with C++, the line after the colon and before the opening brace (shown below), @verbatim : Ipv4Network (network, mask, baseAddress) { @end verbatim Passes the appropriate parameters to the constructor of the base class @code{Ipv4Network}. There are two new calls that we haven't seen immediately after this initialization. They are: @verbatim Ipv4AddressGenerator::SeedNetwork (mask, network); Ipv4AddressGenerator::SeedAddress (mask, baseAddress); @end verbatim We provide an IP address generator class to allow us to programatically allocate IP addresses. The first call to @code{SeedNetwork} gives the address generator a starting network number to use when generating addresses. The second call to @code{SeedAddress} gives the address generator a starting IP address to use. There is a starting network and starting address for each of the 32 possible network masks. Later in the for loop, you will see a call to @code{AllocateAddress} in which the IP address for each node created in the loop is actually generated. The only unfamiliar call in the reset of the constructor will be: @verbatim m_nodes.push_back (node); @end verbatim This is the STL code to add the newly created node to the vector of nodes attached to the bus. For your convenience, we reproduce the entire bus network implementation below: @verbatim /* -*- Mode: C++; c-file-style: "gnu"; indent-tabs-mode:nil; -*- */ /* * Copyright (c) 2007 University of Washington * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 as * published by the Free Software Foundation; * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include "ns3/mac48-address.h" #include "ns3/csma-net-device.h" #include "ns3/csma-topology.h" #include "ns3/csma-ipv4-topology.h" #include "ipv4-bus-network.h" #include "ipv4-address-generator.h" namespace ns3 { Ipv4Network::Ipv4Network ( Ipv4Address network, Ipv4Mask mask, Ipv4Address address) : m_network (network), m_mask (mask), m_baseAddress (address) { } Ipv4Network::~Ipv4Network () { } Ipv4BusNetwork::Ipv4BusNetwork ( Ipv4Address network, Ipv4Mask mask, Ipv4Address baseAddress, DataRate bps, Time delay, uint32_t n) : Ipv4Network (network, mask, baseAddress) { Ipv4AddressGenerator::SeedNetwork (mask, network); Ipv4AddressGenerator::SeedAddress (mask, baseAddress); m_channel = CsmaTopology::CreateCsmaChannel (bps, delay); for (uint32_t i = 0; i < n; ++i) { Ptr node = CreateObject (); uint32_t nd = CsmaIpv4Topology::AddIpv4CsmaNetDevice (node, m_channel, Mac48Address::Allocate ()); Ipv4Address address = Ipv4AddressGenerator::AllocateAddress (mask, network); CsmaIpv4Topology::AddIpv4Address (node, nd, address, mask); m_nodes.push_back (node); } } Ipv4BusNetwork::~Ipv4BusNetwork () { } Ptr Ipv4BusNetwork::GetNode (uint32_t n) { return m_nodes[n]; } }; // namespace ns3 @end verbatim @section Using Ipv4BusNetwork If all you ever want to do with a bus network can be captured in a topology with four nodes on the bus, the preceeding section may seem like a colossal waste of time. This is probably not the case, though. Now that we have a relatively abstract bus class, we can create bus networks with 4, 40 or 4000 nodes with no additional effort. A use of the bus network class is shown in the file @code{bus-netowrk.cc} located in the @code{tutorial} directory. The interesting code is, @verbatim Ipv4BusNetwork bus ("10.1.0.0", "255.255.0.0", "0.0.0.3", DataRate(10000000), MilliSeconds(20), 10); @end verbatim Here we create a bus network with the network number ``10.1.0.0'' and the network mask ``255.255.0.0'' that completes the IP network definition. You can consider these together as ``10.1.0.0/16'' if you prefer. The next parameter tells the bus to start numbering IP addresses of contained nodes at ``10.1.0.3'' (remember the network number will be combined). We provided a data rate of 10 megabits per second and a latency of 20 milliseconds. Finally, we ask the @code{Ipv4BusNetwork} object to create ten nodes in the network. If you are feeling brave, go ahead and change the number of nodes to be 100, 1000, 10,000 or more to generate larger and larger networks. Before you go too far, remember that a trace file will be generated when you run your resulting program and ee asked the trace facility to trace all net device receive events. This will include the reception of the broadcast ARP request by all of the nodes in the simulation, so this can add up quickly. @c ======================================================================== @c Summary @c ======================================================================== @node Summary @chapter Summary This concludes the first part of the tutorial. We have focused on using the @command{ns-3} system to construct various network topologies and to simulate sendng data across the networks; and we've shown you how to use the trace facility to get access to simulation results. We now encourage you to play with the system a little. Experiment with what we have provided. Build a hierarchical network simulation. Perhaps exercise your object design skills and create a new @code{Ipv4DumbbellNetwork} class to create dumbbell networks using the Ipv4BusNetwork class we just created. Hint: An Ipv4DumbbellNetwork @emph{has} two @code{Ipv4BusNetwork} objects; a left side and a right side. In the next part of the tutorial we are going to drop down a level and begin examining the lower levels of the system in more detail. We are going to explain how to change the behavior of the system and eventually how to write new models and applications. This is a good time to make sure that you thorougly understand what we've gone over so far. @c ======================================================================== @c Object Model @c ======================================================================== @node Object-Model @chapter Object Model @cindex Object Model There are two distinctly different meanings associated with the term Object Model. The first speaks to the implementation of an object system --- a system view; and the second speaks to the application programming interface (classes or objects) one uses to access some service or system --- an application view. As an example of the system view sense of the term, the C++ language has an associated object model that describes how objects are laid out in memory, how virtual functions work, how inheritance is implemented, constructor and destructor execution ordering, template instantiation, etc. @cindex API @cindex DOM @cindex Document Object Model In the case of the application view, the Document Object Model is a good example. In the words of W3C, the Document Object Model (DOM) is an application programming interface (API) for HTML and XML documents. It defines the logical structure of documents and the way a document is accessed and manipulated. @cindex API @cindex COM @cindex Component Object Model The Component Object Model (COM) from Microsoft actually spans both meanings of the term and extends further into policy statements. From a system perspective, COM specifies an interface definition language, the layout of objects virtual function tables, the formats of Globally Unique Identifiers and also specifies lifetime management mechanisms for objects via reference counting. From the point of view of the API, COM specifies a number of Interfaces as well as functions such as CoCreateInstance and various threading models. The COM specification extends to policy by disallowing implementation inheritance. @cindex Feynman The @command{ns-3} object model takes the C++ language (system level) object model as its basis, and extends that model by providing an API for software componentry. You may find terms like Component, Interface and QueryInterface in the following discussion, or used informally in other discussions about @command{ns-3}. It is important to understand from the outset that this is the @command{ns-3} object model, and not any other object model. Richard Feynman (an American physicist) once described the behavior of matter and light on a very small scale in the following way, @quotation ``They do not behave like waves, they do not behave like particles, they do not behave like clouds, or billiard balls, or weights on springs, or like anything that you have ever seen.'' @end quotation Just as students of quantum mechanics must rid themselves of preconceptions regarding the behavior of matter at small scales, you should rid yourself of any preconceptions you may have about components, interfaces and APIs for software componentry before continuing. To paraphrase Feynman, @command{ns-3} components do not behave like COM Components, or Java Beans, or CORBA objects, or clouds or weights on springs, or like anything that you have ever seen --- they are @command{ns-3} components. @section The C++ Object Model is the Root of all Things @command{Ns-3} is primarily a C++ system. The system is written in C++ and one can use standard C++ mechanisms for creating and using ns-3 objects. We do not change this at all, nor do we make any pronouncements about the superiority of one mechanism or another. What we will do is provide convenience functions that we think will make creating and managing simulation objects easier. @cindex CreateObject Previously, you have seen objects created using the template function @code{CreateObject} as in the following example: @verbatim Ptr n0 = CreateObject (); @end verbatim This line of code, while it may be unfamiliar to some, is pure C++. If you were to look in the header file ptr.h, you would find the following definition of the @code{CreateObject} template. @verbatim template Ptr CreateObject (void) { Ptr p = Ptr (new T (), false); p->SetTypeId (T::GetTypeId ()); return p; } @end verbatim @cindex template As you can see, this template creates objects of type @code{T} using the operator @code{new}. Its a little harder to find the corresponding delete --- it's in the file @code{object.cc} inside the method @code{Object::MaybeDelete}, but when that @code{Ptr} which you see above goes out of scope it will call @code{Unref} and ultimately the C++ @code{delete} operator will be called. @cindex new @cindex delete The ns-3 system uses the C++ @code{new} and @code{delete} operators, so there is really no reason that you as a user of the ns-3 system are forbidden from using these or any other C++ mechanism. If you so desire, you can take on the responsibility for managing object lifetime (i.e., do not use the @code{Ptr} smart pointer), work directly with the @code{new} and @code{delete} operators and call methods like any C++ object as in the following example: @verbatim MyClass *obj = new MyClass (); obj->Method(); delete obj; @end verbatim @cindex model You, as a competent model author, are encouraged to use whatever methods you think are appropriate in your private code. Remember, however, that the public ns-3 APIs do use smart pointers to pass objects around in an effort to reduce the burden of object lifetime management. If you do intend to export an API publicly, you should use the same object lifetime management approaches as those found in the ns-3 public API if only for consistency. These APIs are there for convenience and consistency, but do not change the fact that in ns-3 all of the objects are really just C++ objects, ultimately created using the C++ new operator with C++ constructor semantics and are ultimately deleted using the C++ delete operator, following C++ destructor semantics. Although it may sometimes appear so, there is really no system- level magic going on in ns-3. Ns-3 components and interfaces are C++ objects just like any other object and our object model is simply a collection of APIs built on the normal C++ object model. @cindex Interface @cindex Abstract Data Type @cindex ADT @cindex Abstract Base Class @cindex ABC @section Interface There are many different ideas floating around of what exactly the term @emph{interface} means. Originally an interface just meant a communication boundary between two entities. As the concepts of object oriented programming (OOP) were surfacing in the 1980s, the term interface was applied to the collection of access methods for the modular entities that were being defined. @cindex OOP @cindex Object Oriented Programming Two distinct approaches developed regarding specifying access mechanisms for objects. The OOP purists were very concerned about object reuse and were led to Abstract Data Types (ADT). These were eventually implemented in the case of C++, as pure virtual methods in Abstract Base Classes (ABC). Another group of folks was more interested in simply specifying object access methods in one place and using inheritance as the primary reuse mechanism. Bjarne Stroustroup, the creator of C++, embraced both approaches. He makes the following interesting observation: @quotation ``Many classes [@dots{}] are useful both as themselves and also as bases for derived classes. [@dots{}] Some classes, such as class @strong{Shape}, represent abstract concepts for which objects cannot exist.'' @end quotation @cindex PIMPL @command{Ns-3} does not pick and enforce a particular approach. In @command{ns-3} an interface is determined completely by a class declaration just as any C++ object interface is declared. If you think of an object as an abstract concept that should be implemented by derived classes, by all means, use the Abstract Base Class approach to interface declaration. If you think that an object should be completely concrete and you foresee no need to ever modify its behavior, feel free to avoid declaring any methods virtual. If you think that an object could be useful as a base class, feel free to declare its methods virtual. If you like to use the PIMPL idiom, again, feel free. If you want to use any combination of these techniques, feel free. We make no restrictions. @cindex API When we speak of an ns-3 interface, we do not worry about interface definition languages, or pure virtual classes, or registries we just think about C++ object declarations and their associated methods. We tend to think of interfaces to objects as simply a private or public API. When we instantiate an @command{ns-3} interface, it is the C++ object model that dictates how that object is brought into existence. When a method is called on an @command{ns-3} Interface, it is the C++ object model that dictates how that method is dispatched. We do, however, provide a base class that endows vanilla C++ objects with capabilities that can be seen as conceptually similar to those provided by Microsoft Component Model @emph{Interfaces}. @section The Ns-3 Object and GetObject @cindex Component Object Model One thing that Microsoft arguably got right in the Component Object Model was the idea of Interface aggregation and discovery via QueryInterface. We have embraced these ideas in @command{ns-3}. This was done primarily to address a common problem in large software systems. A good example of this problem happens in the @command{ns-3} Node class. @cindex OOP @cindex weak base class @cindex base class bloat @cindex Swiss Army Knife class @cindex Node If one were to take the standard OOP view of specializing a @code{Node} into an internet host, for example, one would typically inherit from the @code{Node} base class and include functionality to implement such things as internet routing and a TCP/IP protocol stack. Other types of @code{Node}s might inherit from the node class and specialize in different ways, or further specialize the internet host class, treating it as a base class. This can result in a complicated inheritance tree in which some specializations are simply not available to other branches of the tree which can make reuse difficult or impossible. This is known as the @emph{weak base class} problem and creates pressure to drive functionality up the inheritance tree into the base classes. This, in turn, results in @emph{base class bloat} and the resulting @emph{swiss army knife} base classes which end up trying to do everything in one place. Even if one successfully avoided these swiss army knife base classes, one would also want to be able to treat new specializations of @code{Node} generically in the system. This means one would pass references to the base class (@code{Node}) across public APIs. This introduces @emph{upcasts} prior to passing across public APIs and corresponding @emph{downcasts} on the other side in order to gain access to required specialized functions. As the inheritance tree becomes more complicated, this approach can cause another related problem known as the @emph{fragile base class} problem. This happens when changes to the base class cause unexpected problems in the various and sundry subclasses. These effects seem always to result in a positive feedback loop driving everything into the base class and destroying much of the encapsulation which is a hallmark of the object oriented approach. @subsection Interface Composition @cindex Node There is a completely different way to address the Node specialization problem. Instead of approaching the situation using inheritance, one can look at the problem as one of composition. We can look at the @code{Node} class as a container of sorts that holds other objects. In this case, the objects would be instances of the classes implementing the internetwork routing code, or the TCP/IP protocol stack described above. This approach preserves the encapsulation and solves the weak base class, base class bloat and fragile base class problems; but the question of method dispatch immediately comes to mind. @cindex delegation In many systems, @emph{delegation} is used. The base class, @code{Node}, in this approach would provide methods that simply forward to the objects implementing the desired functionality. This situation clearly does not address the base class bloat problem since dispatch methods must be added to the base class. The situation is mitigated somewhat by pushing the implementation of the dispatch methods to contained objects, but the fundamental problems are still present. What is really needed is a way to compose objects but at the same time keep the interfaces to those objects separated. @cindex aggregation Composition, usually called @emph{aggregation}, along with runtime Interface discovery is the solution that Microsoft originally championed and that @command{ns-3} has adopted --- albeit with many simplifications and a few name changes. @subsection Objects and Interfaces @cindex COM @cindex QueryInterface Now that we have mentioned Microsoft COM and are almost obligated to mention the terms Interface and QueryInterface. For those familiar with COM, loosely speaking, QueryInterface is to COM as GetObject is to @command{ns-3}. The analogy, while good conceptually, is superficial from an implementation point of view. @cindex Node Addressing our current example of a @code{Node}, generically speaking, each node needs to aggregate an object that will implement internetwork routing and TCP/IP. The system will need to provide a mechanism for locating the aggregated objects and allow a client to discover them. @cindex aggregation @cindex Object These aggregated objects have interfaces in the C++ sense of collections of method signatures. In @command{ns-3}, when objects are capable of participating in this aggregation process, they are called @command{ns-3} @code{Objects}. @code{Objects} receive the functionality required for this participation by inheriting from the @command{ns-3} base class @code{Object}. Note well that when we write the word @code{Object} (note the uppercase 'O' in the spelling and the change of font) we are referring to a kind of C++ object that has inherited the capability of participating in an aggregation. The @command{ns-3}-specific word @code{Object} can have a significantly different meaning than that of a vanilla C++ object outside the aforementioned inheritance tree, and the difference is only readily apparent via context. In this tutorial we will always write the @command{ns-3}-specific kind of @code{Object} in a fixed font; and will write the vanilla C++ term object in normal font. In conversation, you will need to be careful to understand which term is meant: object or @code{Object}. Once an object has inherited from class @code{Object} it has the ability to @emph{host} an aggregation. This means that it has the ability to add other @code{Objects} to its aggregation via the method @code{AggregateObject}. It also means that it can provide a service to @emph{discover} other objects in its aggregation via the method @code{GetObject}. @cindex base class Technically, the class named @code{Object} is simply a base class that you will inherit from if you want your @code{Objects} to support aggregation and discovery. Many systems have a base class that implements common functionality and these base classes are typically called somthing like Object. The @command{ns-3} version of this base class relates primarily to @code{Object} aggregation and discovery, although it does also provide methods to help with intrusive reference counting and tracing as well. When a C++ object inherits from the ns-3 Object base class, it is conceptually promoted to an ns-3 @code{Object} irrespective of how the object was declared (e.g., as an abstract base class, concrete class, with virtual methods, etc.). In ns-3, you should associate inheritance from the class named @code{Object} with promotion of an object to the status of some locatable @code{Object} rather than with the form of the class declaration. @cindex COM @cindex CORBA @cindex ORBit For those of you unfamiliar with Microsoft COM, CORBA or ORBit, this might sound obvious. For those of with such a background, the point we are making is that there is no such thing in @command{ns-3} as a separate Interface declaration, no such thing as an Interface Definiition Language, no such thing as a UUID or GUID, etc. In @command{ns-3} we just work with C++ objects that may be given some very useful abilities by inheriting from the @command{ns-3} base class @code{Object}. @command{Ns-3} @code{Objects} are not required to inherit from classes composed of pure virtual methods in order to define an Interface. It's all really just ``plain old C++.'' To summarize, when you instantiate an object that inherits from the @code{Object} class, you will have a C++ object that has four important properties: @cindex AggregateObject @cindex GetObject @itemize @bullet @item The @code{Object} has a C++ interface defined by the collection of method signatures in its inheritance tree; @item The @code{Object} has some way to identify its underlying class uniquely; @item The @code{Object} is a kind of container that has the ability to aggregate other @code{Objects} using the method @code{AggregateObject}; @item The @code{Object} exports a method called @code{GetObject} that allows for discovery of other aggregated @code{Objects}. @end itemize @cindex base class @cindex Object It is crucially important to understand what we have described here (especially for those coming from other systems that provide similar functionality). A given C++ class has an object access interface that is essentially the collection of method signatures specified in its inheritance tree. This is a C++ object model thing. Ns-3 provides a base class from which the class in question can inherit and be promoted to the status of @code{Object}. Once a class becomes an @code{Object} it has inherited the ability to aggregate and search for other @code{Objects} that are added to its aggregation. That last detail is important. In @command{ns-3} @code{Objects} are both containers and specifications for a object method access. We have previously mentioned that the @code{Node} class acts as a container. In fact, the @code{Node} class inherits from @code{Object} and is itself an @command{ns-3} @code{Object}. So, when the @code{Node} object is created it is really an aggregation of one @code{Object} and you can call @code{AggregateObject} or @code{GetObject} on the resulting @code{Node} object. Along with being an aggregation, the @code{Node} class also describes a public interface. THis public interface (API) is declared just as any C++ object is declared, via its class methods as specified in the inheritance tree. For those steeped in COM or CORBA, this is where the concept of Interface works in @command{ns-3}. Remember that it is generally true that @code{Objects} are both aggregations and APIs. @subsection Aggregations @cindex aggregate The figure below shows how an @code{Object} could be illustrated in detail. The line with the circle at the top of the diagram represents the appearance of the @code{Object} API to the external world. This circle and line are together called a lollipop because of its superficial similarity to a kind of childs candy. @sp 1 @center @image{oneobj,,,,png} @cindex API You could declare this API and associated @code{Object} quite simply using a non-virtual class as follows, @verbatim class A : public Object { public: static ns3::TypeId GetTypeId (void) { static ns3::TypeId tid = ns3::TypeId ("A") .SetParent (Object::GetTypeId ()) .AddConstructor (); return tid; } A () { } void MethodA (void); }; @end verbatim The methods that are then available via the API labeled @code{A} in the figure above are the methods inherited from the @code{Object} base class (@code{GetObject}, @code{Ref}, and @code{Unref}) and those from class @code{A} (@code{MethodA}). Note that you must declare a @code{TypeId} in your @code{Object} class, and it must be declared static to make it class-wide in scope. This @code{TypeId} is a unifying element in the @command{ns-3} object model and uniquely identifies @code{Objects} at run-time as being instantiated from a particular class. We'll have much more to say about @code{TypiId} shortly. You can think of the arc and arrow device coming off each side of the illustrated @code{Objects} as part of a connector. These connectors allow @code{GetObject} to search aggregations for an instance of a class type. The figure below shows an aggregation of three @code{Objects}: A, B and C. The class declarations for classes @code{B} and @code{C} are substantially similar to that of class @code{A}. @sp 1 @center @image{threeobj,,,,png} You can visualize these @code{Objects} as being snapped together like Lego building blocks if you like. When @code{Objects} are aggregated, a @code{GetObject} search path is formed through the connectors. In order to create this aggregation you will first need to create the @code{Objects}. These are just normal, everyday C++ objects that we can create using the @code{CreateObject} template function and manage using smart pointers. The following code should be obvious to you by now: @verbatim Ptr a = CreateObject (); Ptr b = CreateObject (); Ptr c = CreateObject (); @end verbatim @cindex aggregation When you create an aggregation, you pick one of the @code{Objects} of the aggregation to think of as the container. In this case well pick @code{Object} A. In order to aggregate an @code{Object}, you simply call the method @code{AggregateObject} that your class has inherited from class @code{Object}. The following code will aggregate @code{Object B} and @code{Object C} onto the @code{Object} (and container/aggregation) @code{A}. @cindex AggregateObject @cindex GetObject @cindex Object @verbatim a->AggregateObject (b); a->AggregateObject (c); @end verbatim Thats all there is to it. Now that you have those connectors snapped together, you can ask each of the @code{Objects} in the aggregation for any of the other @code{Objects} in the aggregation. Lets look at a simple example: @verbatim Ptr newB = a->GetObject (); @end verbatim Now, the explanation of what this snippet does is not as simple as writing it. The left hand side of this assignment declares a smart pointer to the class @code{B} to help with memory management of the returned @code{Object} pointer. You should be very familiar with smart pointers at this stage of the tutorial. The right hand side illustrates how @code{GetObject} is acutally used. The method @code{GetObject} is templated. The assocated template parameter (between the brackets) specifies the @emph{class} that is being requested. This is important. Since it is the class type that specifies the search criteron, there can be only one instance of a particular class present in an aggregation. Looking back a little, although the parameter to @code{AggregateObject} appears to be a vanilla C++ object (@code{b} or @code{c} above), it actually represents (is an instance of) a class that has an associated @code{TypeId} and inherits from @code{Object}. When you call @code{GetObject} you specify the search criterion (using the template parameter) as a class name. This referenced class must also have an associated @code{TypeId} and must also have inherited from @code{Object}. This may be summarized by saying that @code{AggregateObject} takes an @emph{instance} of an object of a particular class that inherits from @code{Object}. GetObject looks for a @emph{class} of a particular type (that again inherits from @code{Object}) and possibly returns an aggregated object instance of that type. Now that you have those conceptual connectors snapped together, you can ask each of the @code{Objects} in the aggregation for any of the @code{Objects} in the aggregation. For example we could walk the @code{Objects} asking each for the next in the aggregation. First we would ask the @code{Object} pointed to by the smart pointer @code{a} to look for the @code{Object} @code{class B}: @verbatim Ptr newB = a->GetObject (); @end verbatim Next, we can ask the @code{Object} pointed to by the smart pointer @code{newB} to look for the @code{Object} representing @code{class C}: @verbatim Ptr newC = newB->GetObject (); @end verbatim @cindex Object Then, we can ask the @code{Object} pointed to by the smart pointer @code{newC} to look for the @code{Object} representing @code{class A} and complete our circuit of the aggregation: @verbatim Ptr newA = newC->GetObject (); @end verbatim @cindex GetObject @code{GetObject} has some important properties that we need to go over. Technically, @code{GetObject} is a @emph{symmetric}, @emph{reflexive} and @emph{transitive} operation with respect to the set of aggregated @code{Objects}. @subsubsection Symmetry @cindex symmetry The symmetric nature of @code{GetObject} guarantees that if one performs a @code{GetObject} on a given @code{Object} for the class of that same @code{Object}, that @code{GetObject} must succeed. In other words, the fact that you accessed the aggregation via an instance of an @code{Object A} in the aggregation implies the reachability of that @code{Object} in the aggregation. This is usually written (by Microsoft) as, @center must succeed (A >> A) We can illustrate this property with the code snippet, @verbatim Ptr symmetricA = a->GetObject (); NS_ASSERT (symmetricA); @end verbatim Here we take as given an interface (smart) pointer --- named @code{a} --- on which we perform a @code{GetObject} looking for the class that represents that same @code{Object}. This call must always succeed and a smart pointer to the aggregated instance of that class is returned. @subsubsection Reflexivity @cindex reflexivity Calls to @code{GetObject} must also be reflexive. This means that if you successfully @code{GetObject} for @code{Object B} from @code{Object A}, then you must always be able to @code{GetObject} for @code{A} from @code{B}. This is usually written as, @center must succeed (A >> B, then B >> A) This property can be illustrated with the code snippet, @verbatim Ptr b = a->GetObject (); Ptr reflexiveA = b->GetObject (); NS_ASSERT (reflexiveA); @end verbatim If the first @code{GetObject} on @code{Object A} looking for @code{Object B} succeeds, then a @code{GetObject} on @code{Object B} looking @code{Object A} must succeed. @subsubsection Transitivity @cindex transitivity @code{GetObject} must also be transitive. This means that if one can find @code{Object B} from @code{Object A}, and @code{Object C} from @code{Object B}, then one must also be able to find @code{Object C} from @code{Object A}. This is usually written as, @center must succeed (A >> B, and B >> C, then A >> C) This property can be illustrated with the code snippet, @verbatim Ptr b = a->GetObject (); Ptr c = b->GetObject (); Ptr transitiveC = a->GetObject (); NS_ASSERT (transitiveC); @end verbatim If you can get to @code{Object B} from @code{Object A}, and you can get to @code{Object C} from @code{Object B}, then a @code{GetObject} on @code{Object A} looking for @code{Object C} must also succeed. @subsection Creating the TypeId @cindex TypeId @cindex GetTypeId The final piece of this puzzle is the @code{TypeId}. Recall that the declaration our eample object above included the following code @verbatim static ns3::TypeId GetTypeId (void) { static ns3::TypeId tid = ns3::TypeId ("A") .SetParent (Object::GetTypeId ()) .AddConstructor (); return tid; } @end verbatim This is the bit of code that ties this all together. For those unfamiliar with the idioms involved, this declaration can be rather dense. First, let's examine the function declaration itself. The following code, @verbatim static ns3::TypeId GetTypeId (void) ... @end verbatim declares a function that will be associated with all of the instances of the given class. This is a function, not a method, in that it can be accessed without a @emph{this} pointer; but it is associated with the class in a namespace sense. The use of this kind of declaration allows one to write, @verbatim return A::GetTypeId (void); @end verbatim if the @code{TypeId} is needed for our @code{class A}. More generically the class name can be substituted in a template, as is done deep in the @command{ns-3} object system. From this perspective, if you leave out the middle of the function definition, the boundaries should make sense to you. @verbatim static ns3::TypeId GetTypeId (void) { return tid; } @end verbatim @cindex function-local variable You are obviously looking at a global function associated with your class that simply returns a @code{TypeId}. Now, what about the rest. The code @verbatim static ns3::TypeId tid = ns3::TypeId ("A") .SetParent (Object::GetTypeId ()) .AddConstructor (); @end verbatim when found inside the function declaration is called a function-local variable with associated initialization. It'll be easier to pick this statement apart piece by piece as well. The first line, @verbatim static ns3::TypeId tid = ... @end verbatim is the declaration of the function-local variable tid. This is essentially an initialized global variable, the scope of which has been reduced to within the enclosing method. You can think of this as a kind of global variable that can only be accessed right there where it is created. If the variable is initialized, this amounts to the same behavior as if a global static initializer was declared in a namespace of the same name as your class. Global static initializers are guaranteed by the C++ language definition to be executed before your main procedure is entered. So are function-local variables. The variable that is being initialized is of type @code{ns3::TypeId}, is named @code{A::tid} since it is inside the class declaration for @code{class A}, and is initialized by a call to the constructor for the class @code{TypeId}. The constructor for @code{TypeId} takes a @code{std::string} that can be used to locate the type information for your class. We usually privide the class name as the string. Hopefully, this much of the declaration is now clear: @verbatim static ns3::TypeId GetTypeId (void) { static ns3::TypeId tid = ns3::TypeId ("A") ... return tid; } @end verbatim All that is left now are the lines including @code{SetParent} and @code{AddConstructor}. @verbatim static ns3::TypeId tid = ns3::TypeId ("A") .SetParent (Object::GetTypeId ()) .AddConstructor (); @end verbatim The last bit may seem quite odd at first glance, but don't let the way the code is broken up over several lines throw you. If you saw something like, @verbatim pointer->TypeId()->SetParent()->AddConstructor(); @end verbatim you probably wouldn't hesitate at all. Clearly, you would think, a method called @code{TypeId} is called using the pointer called @code{pointer} as shown below. @verbatim pointer->TypeId() @end verbatim The method @code{TypeId} must further return a pointer to an object that has a method called @code{SetParent}. Just as clearly, @code{SetParent} must return a pointer to an object that has a method called @code{AddConstructor}. The same sort of thing is happening in our code snipped, except we are using references instead of pointers. Perhaps if we rearrange this code to live on one line it will be clearer. @verbatim ns3::TypeId ("A").SetParent (Object::GetTypeId ()).AddConstructor (); @end verbatim It's just a string of method calls. The remaining question is then, what do those three methods do. The first, @code{ns3::TypeId ("A")}, simply allocates a new type in the system and allows you to refer to it in the future by a string. We have mentioned inheritance trees often in the previous discussion. The second method, @code{SetParent} associates the class being defined with its parents in the tree. Finally, the @code{AddConstructor} method allows you to specify a constructor to be used when an instance of your class is created using @code{CreateObject}. @verbatim AddConstructor (); @end verbatim You can interpret this as explaining to the @command{ns-3} object ssytem that you have a constructor named @code{A::A} which takes no parameters. You are saying that this constructor should be used when @code{CreateObject} is called with no parameters. By including the structure of the inheritance tree, in @command{ns-3} we can use implementation inheritance to easily create new @code{Objects}. You are prevented from doing so in Microsoft COM, but this was almost universally identified as a problem. So, looking at the entire @code{GetTypeId} declaration again, @verbatim static ns3::TypeId GetTypeId (void) { static ns3::TypeId tid = ns3::TypeId ("A") .SetParent (Object::GetTypeId ()) .AddConstructor (); return tid; } @end verbatim it should be clear what is happening. @subsection A Very Real Example @cindex Node @cindex AggregateObject @cindex GetObject @cindex Object At this point you may be asking yourself what the point of all of this is, since you already had those pointers laying around when you created the objects. The typical case is that one will create and aggregate some number of @code{Objects} in a constructor and return only a pointer to a single @code{Object} as in our canonical example with @code{class Node}. In this case, the @code{Node} would be created and the @code{Node} constructor might create and call @code{AggregateObject} to aggregate the @code{Objects} for internetwork routing and TCP/IP. From an external point of view, these aggregated objects may be discovered at run-time using @code{GetObject}. Generally one tends to think of one of the @code{Objects} in the aggregation as being the container and other @code{Objects} being aggregated to that container. In the case of a Node, for example, it is quite natural to think of the Node as being the container which contains protocol stacks, internet routing, etc. So, lets start thinking about a real example by calling the container @code{Object Node} instead of @code{A} as we have been. The creation of this @code is found all over our example programs. For example, you will find code like the following in @code{samples/simple-point-to-point.cc}: @verbatim Ptr n = CreateObject (); @end verbatim It may appear obvious to you now that the @code{InternetNode} class name provided to the template function @code{CreateObject} means that @code{InternetNode} is an @command{ns-3} @code{Object} and you will be able to call @code{GetObject} on the resulting smart pointer. Well, I'm afraid that's not entirely true. It's slightly more complicated. Take a look at @code{src/internet-node/internet-node.h} and find the class declaration for @code{InternetNode}. @verbatim class InternetNode : public Node { public: InternetNode(); ... }; @end verbatim @cindex GetTypeId @cindex TypeId @cindex Object There is no declaration of a @code{static TypeId GetTypeId (void)} in this class. This means that the @code{InternetNode} is really not an @code{Object} for which you can @code{GetObject}. It turns out that the @code{InternetNode} is an @emph{implementation class} of the @code{Node Object}. You may recall that there can be an implicit cast in a smart pointer assignment if the cast is to a visible, unambiguous base class. That is, in fact, what is happening here. Now, take a look at @code{src/node/node.h} and find the class declaration for @code{class Node}. There you will find, @verbatim class Node : public Object { public: static TypeId GetTypeId (void); ... }; @end verbatim Class @code{InternetNode} inherits from class @code{Node} that, in turn, inherits from class @code{Object}. It is @code{Node} that provides a @code{GetTypeId} method. Therefore it is @code{Node} that is an @command{ns-3} @code{Object}. Note well that @code{InternetNode} is not an @code{Object} in the sense that one should call @code{GetObject} on an aggregation looking for an @code{InternetNode} class. That is, you should not do, @verbatim Ptr i = node->GetObject (); @end verbatim since there really is not InternetNode::GetTypeId. It is @code{Node} that is the @emph{proper} @code{Object} in this case and you should view @code{InternetNode} as an implementation of the @code{Node Object}. This may become clearer as we look a little deeper. We spoke of a protocol stack that is aggregated to a @code{Node} in our discussions above, what we see in the real @command{ns-3} code is that this is represented by the @code{Ipv4 Object}. If you look in @code{src/node/ipv4.h} you will find, @verbatim class Ipv4 : public Object { public: static TypeId GetTypeId (void); ... }; @end verbatim Since class @code{Ipv4} inherits from class @code{Object} and has a @code{GetTypeId}, it is an @command{ns-3} @code{Object}. If you look in @code{src/node/ipv4.cc} you will find, @verbatim TypeId Ipv4::GetTypeId (void) { static TypeId tid = TypeId ("Ipv4") .SetParent (); return tid; } @end verbatim After all of this reading you know that this code snippet is asking the system to create a unique @code{TypeId} for the @code{Ipv4} class and declares that @code{Ipv4} inherits from class @code{Object}. This is what makes an @code{Ipv4} an @code{Object}. @cindex Ipv4 It turns out that the Ipv4 class is an abstract base class (ABC). There are a number of pure virtual methods declared in that class. This means that an @code{Ipv4} object may not be instantiated. This is reflected by the fact that there are no constructors registered in the @code{GetTypeId} method above. What is instantiated in the real system is an implementation class, called @code{Ipv4Impl}. This class inherits from @code{Ipv4} and provides the required virtual methods. This is where understanding what is an @code{Object} and what is not can get tricky. The @code{Object} is the @code{Ipv4} class since that is where the @code{GetTypeId} is found. The fact that you see @code{GetTypeId} there tells you that the @code{Ipv4} class is the class for which you can @code{GetObject}. @cindex implementation class The class @code{Ipv4Impl} provides an implementation for the pure virtual methods in @code{Ipv4}. Since class @code{Ipv4} cannot be instantiated, one instantiates the @code{Ipv4Impl} class to create an @code{Ipv4} @code{Object}. You will use the @code{CreateObject} template function to create an object that implements the methods of an @code{Object}. You can probably see how this gets even more tricky in conversation. Once the @code{Ipv4Impl} object is instantiated, the resulting pointer is immediately cast to an @code{Ipv4} pointer. Clients will then use the methods specified in the @code{Ipv4} class to access the @code{Ipv4 Object} methods which are, in turn, implemented in the @code{Ipv4Impl} object. If you now look in the file, @code{src/internet-node/internet-node.cc} you will see the following code in @code{InternetNode::Construct} that creates the @code{Ipv4} Interface and aggregates it. @verbatim Ptr ipv4Impl = CreateObject (ipv4); ... Object::AggregateObject (ipv4Impl); @end verbatim Note that the parameter @code{ipv4} passed to the @code{CreateObject} template function is actually a pointer to an @code{Ipv4L3Protocol} which you can ignore at this point --- it doesn't really have anything to do with the @code{Ipv4} Interface. This is exactly the same thing that is happening in the case of the @code{InternetNode}. @verbatim Ptr n = CreateObject (); @end verbatim @cindex implementation object @code{CreateObject} is being called to create an implementation object, in this case @code{InternetNode}, which implements the methods of the @code{Node Object}. It is the resulting @code{Node Object} which you would use as the container and it is the @code{Node} class that you would use as the template parameter when calling @code{GetObject}. In the same way, you would @emph{not} want to do, @verbatim Ptr ipv4 = node->GetObject (); @end verbatim Rather you should understand that the @emph{proper} @code{Object} is the @code{Ipv4} not the @code{Ipv4Impl} and do the following, @verbatim Ptr ipv4 = node->GetObject (); @end verbatim @cindex CreateObject This does illustrate that the fact that whether an object created by @code{CreateObject} is or is not an @code{Object} in the usual sense can be quite well hidden if you are casually looking at the object creation code. The designers of the system had long and involved discussions on this issue and in the end decided that mnemonic aids such as Hungarian notation were a stylistic thing and you should just refer to the system documentation to determine what objects are @command{ns-3} @code{Objects} and what the APIs of those @code{Objects} actually are (RTFM --- as in Read the Fine Manual, of course). @cindex AggregateObject @cindex Object In the case of @code{Ipv4Impl}, you know that the class inherits somehow from @code{Object} since there is a call to @code{AggregateObject} that refers to an instance of an @code{Ipv4Impl}. You will have to go to the header file @code{src/internet-node/ipv4-impl.h} and find that @code{Ipv4Impl} inherits from class @code{Ipv4}. You will then have go to the file @code{src/node/ipv4.h} and see that it inherits from @code{Object} and defines a @code{GetTypeId}. Thus the @code{Object} for which you can @code{GetObject} is really the @code{Ipv4 Object}. Returning to some real @command{ns-3} example code, lets take a look at @code{examples/simple-point-to-point.cc}. You will find the following code in this file: @verbatim Ptr n0 = CreateObject (); ... Ptr ipv4; ipv4 = n0->GetObject (); ipv4->SetDefaultRoute (Ipv4Address (``10.1.1.2''), 1); @end verbatim @cindex InternetNode @cindex Node @cindex Object @cindex GetObject The first line creates an @code{InternetNode} implementation object and casts the resulting smart pointer to a @code{Node} as we have discussed extensively. The next line shown declares a smart pointer to an @code{Ipv4 Object}. We then do a @code{GetObject} on the @code{Node} looking for the @code{Ipv4 Object}. You know since you've read every line of this tutorial in detail exactly how that @code{Ipv4 Object} got into every @code{Node}. You know that the @code{GetObject} will return a smart pointer to its aggregated @code{Ipv4} Interface. Once we have the @code{Ipv4} smart pointer, we simply use it as if it were any other C++ object. The last line shows this by setting the default route for the node. @section Caveats There are a few things that you should remember but which may not be immediately obvious. @subsection Ns-3 Objects are Associated with Classes not C++ objects @cindex Object @cindex GetObject @cindex iterate @cindex aggregation @cindex GetNDevices Okay, you can see some of the problems with the terminology popping up again. We are reminding you that when you do a GetObject you are providing the key to the lookup by giving a class name and not anything that is unique to a C++ object. You cannot add more than one @code{Object} of a given type (class name) to an aggregation. If you need to contain a number of @code{Objects} of the same type in the same aggregation, you will need to provide a separate container over which you can iterate. For example, the @code{Node} class provides methods, @verbatim uint32_t GetNDevices (void) const; Ptr GetDevice (uint32_t index) const; @end verbatim that are used iterate over the multiple @code{NetDevice} @code{Objects} associated with it. @emph{Remember: Object types do not identify objects.} @subsection Dont use GetObject to Check Your Own Type. @cindex GetObject It is tempting to use @code{GetObject} as a form of runtime type information. Dont do it. You have no control over what @emph{other} object may be added to your aggregation. Someone else may have appropriated (reimplemented) your type and aggregated themselves onto the aggregation. Consider a socket factory implementation. Sockets can be either UDP sockets or TCP sockets. A socket factory will have a generic @code{SocketFactory} Object and either a UDP specific interface for setting UDP parameters or a similar TCP-specific interface. Consider what might happen if you declared your socket factory as a partially abstract base class, and then provided separate implementations for UDP and TCP specific methods of this factory in separate concrete classes. Now consider what might happen if you used @code{GetObject} in your base class to determine if you were a UDP or a TCP factory. If a factory, say the UDP version, were not aggregated to any other @code{Object}, the base class could @code{GetObject} on itself for the UDP-specific class name. If the @code{GetObject} succeeded, it could then infer that it was a UDP implementation and would then do any UDP-specific tasks it could. [Experienced C++ folks are cringing about how horrible this design is, but bear with me --- its a simple illustration of a specific and perhaps not-too-obvious problem.] If another factory, say the TCP version, were not aggregated to any other Interface, the base class could @code{GetObject} on itself for the UDP-specific interface. If this failed, it could then infer that it had a TCP implementation and would then do any TCP-specific tasks it could. Now, what happens when these two working objects are aggregated together by some innocent end-user. Since the @code{Objects} are conceptually snapped together, the TCP implementation would suddenly begin finding the UDP Interface from the other class factory and think it was the UPD implementation. @emph{Objects should not be used as run-time type information.} @section Connecting the Dots @cindex Object @cindex GetObject @cindex AggregateObject @cindex GetTypeId @cindex API This may all sound very complicated to you if this is your first exposure to these concepts. It may be annoying if I tell you that its really not as hard as it sounds. Rest assured that if you take some time, look at and understand the examples and write a little test code it will all come together for you. Grep around the system for @code{AggregateObject} and @code{GetObject} and take a look at how we have used them. This will also give you a good idea of what our core @code{Objects} and associated APIs are. If you grep for @code{GetTypeId} you will find most, if not all of the @code{Object} API interface declarations in the system. The more you see this idiom in use, the more comfortable you will be with the idea and the more you will see how this addresses the weak base class, swiss army knife base class, and fragile base class problems I explained at the beginning. As I alluded to earlier, the developers had long discussions regarding how to make navigating the @code{Object} environment easier. The primary issue was how we could make it easier to convey to you, the model writer, that an object was an @code{Object}. We originally used similar terminology as Microsoft COM and used QueryInterface instead of @code{GetObject}. One suggestion was to adopt the convention that classes that implemented Interfaces must begin with the letter I. Microsoft does this, as exemplified by the class IUnknown. We also toyed with the idea of beginning our header files with ``i-'' as in ``i-ipv4.h.'' We considered forcing some structure on Interfaces with a pure virtual class specification, the names of which begin with an I; and corresponding implementations, the names of which begin with a C. This all got out of hand fairly quickly. In the end we decided that we were really discussing issues of programming style, and we really could not come up with a strong reason to impose any particular solution. No matter what direction we took, we ended up with some form of extra confusion or extra complexity somewhere in the system. The resulting system is extremely flexible and easy to use. It is, unfortunately, sometimes hard to document and talk about. @cindex Feynman If it helps you to think in terms of Microsoft COM and Interfaces, by all means do so, just be aware that even though @command{ns-3} @code{Objects} descend from COM in some sense, there are subtle differences that may get you lost or into trouble. So to paraphrase Feynman one more time, @quotation ``@command{Ns-3} @code{Objects} do not behave like COM Components, or Java Beans, or CORBA objects, or clouds or weights on springs, or like anything that you have ever seen --- they are @command{ns-3} components.'' @end quotation Just get very familiar with the @command{ns-3} object model. It is the heart of the system and if you do not understand it you will not understand how to write an @command{ns-3} model properly. @c ======================================================================== @c Doxygen @c ======================================================================== @node The-Doxygen-Documentation-System @chapter The Doxygen Documentation System @node How-To-Change-Things @chapter How to Change Things @node How-To-Set-Default-Values @chapter How to Set Default Values @node How-To-Write-A-New-Application @chapter How to Write a New Application