Tidy Tutorials: October 2009

Double Dispatch and Visitor Pattern Example

Double dispatch is a way to call different (but-related) concrete functions depending on the runtime types of the various objects involved in the call. IT basically replacing static compile time casting with dynamic resolving of calls.

Here is an example of the a double dispatch. Let's say it's a game. You have a wildanimal that attacks types of humans, and a warrior which derives from human. Let us consider a wildanimal attacking a warrior. I strongly suggest running this code for a better understanding.

The problem is on line 60 you have w->attack(h). This leads to static compile time function overloading and you end up selecting the 'Human Attack' at line 44. This leads to a wild animal attacking a human despite the fact that it should be attacking a warrior.

The solution is to have a DD that allows the proper dynamic cast by using the 'this' pointer (line 26). Now when you call h->attackedbywildanimal(w); at line 54 you will get the correct wild animal attacking a warrior instead of a human.

Note that a simpler but far less flexible solution would have been to simply statically cast the pointer at line 50 to be w->attack((warrior*)h);.

Note that to have all the code in a single file I needed class definitions at the top and the wildanimal implementation is actuall at the bottom. Remind me to use java next time!

 1. #include <iostream>
 2. 
 3. using namespace std;
 4. 
 5. class human;
 6. class warrior;
 7. class wildanimal{
 8. public:
 9.     virtual void attack(human*);
10.     virtual void attack(warrior*);
11. };
12. 
13. class human{
14.   public:
15.     virtual void attackedbywildanimal(wildanimal *animal){
16.         animal->attack(this);
17.     }
18.     void GetHealth(){
19.         cout<<"Human health\n";
20.     }
21. };
22. 
23. class warrior : public human{
24. public:
25.     virtual void attackedbywildanimal(wildanimal *animal){
26.         animal->attack(this);
27.     }
28.     void GetHealth(){
29.         cout<<"Warrior health\n";
30.     }
31. };
32. 
33. void wildanimal::attack(human* h){
34.       cout<<"Animal attacks human ";
35.       h->GetHealth();
36. }
37. void wildanimal::attack(warrior* w){
38.       cout<<"Animal attacks warrior ";
39.       w->GetHealth();
40. }
41. 
42. class boar: public wildanimal{
43. public:
44.     void attack(human* h){
45.         cout<<"Boar attacks human ";
46.         h->GetHealth();
47.     }
48.     void attack(warrior* w){
49.         cout<<"Boar attacks warrior ";
50.         w->GetHealth();
51.     }
52. };
53. 
54. int main(int argv, char** argc){
55.   
56.     human * h = new warrior();
57.     wildanimal * w = new boar();
58.   
59.     //Problem
60.     w->attack(h);
61.     
62.     //Double Dispatch solution
63.     h->attackedbywildanimal(w);
64. }

Hide line numbers

Design Patterns Chart - Periodic Table

Design pattern periodic table. The Gang Of four (GoF) design patterns consist of the fundamental creational, structural and behavioural patterns that form the basis for all other patterns. Are these in fact fundamental? Can they be refactored? Let us explore this topic to better understand desgin patterns and their use in everyday programming.

GoF's original books was "Design Patterns: Elements of Reusable Object-Oriented Software" by ErichGamma, RichardHelm, RalphJohnson and John Vlissides.

Creational Patterns

Structural Patterns

Behavioural Patterns

FM Factory Method								Px Proxy	B Bridge
AF Abstract Factory	Pt Prototype	Tm Template	Cd Command	Md Mediator	O Observer	In Interpreter	CR Chain of responsibility	A Adapter	Fy Flyweight
Bu Builder	S Singleton	Sr Strategy	Mm Memento	St State	It Iterator	V Visitor	Cp Composite	D Decorator	Fc Facade

SQL Server C++ ODBC Connection Pool

Aim
The aim of this bit of code is to implement an SQL Server ODBC C++ Database Pool in Windows. A Database pool is normally used by applications to speed up the process of getting a Connection.

If you are just starting with ODBC and would like a simple tutorial first the please go to SQL Server C++ ODBC Example.
If you would like to see this DBPool implemented in MySQL the visit MySQL Connector C++ Database Pool .

Assumptions
This article assumes that you have SQL Server installed and running. For purposes of this example the the project directory is CppDatabasePool. Also it is assumed that you have VC++ installed and configured. See Problems and Solutions Installing VC++ Express Edition, to install and run vcvars32.bat.

Versions used in this example

Sofware/Component	Image
Windows XP SP2	N/A
Visual Studio Express Editions 2008 VC++	N/A

You can download the zipped files here.

CrossHelper.h and CrossHelper.cpp - Define and implement the crossplatform functionality between windows and linux - I haven't finished the linux part yet.

CommonDatabasePool.cpp - This is the primary file that has a template and a map that stores the cached connections and a thread that runs to ping the database connections to keep them alive and prevent an inactive timeout. The idea is to potentially be able to use this template with different database implementations .

sqlserver/SQLSvrPool.h and sqlserver/SQLSvrPool.cpp - These classes actually inherit from the template above and implement the C++ ODBC specific tasks. You will be using this class but not editing it.

examples/ExampleSQLSvr.cpp - This has an example of how you use the pool.

makefile.sqlsvr - Used to build the example. DO NOT forget the "/D WINDOWSXX" flag as this is needed to compile the widows version. Also remember to substitute your local directories for the ones in the bat file. You should link to odbc32.lib binary for compile.

Below is the ExampleSQLSvr.cpp file.

 1. /* Copyright 2009 Righteous Ninja AKA P.S. Ching*/
 2. #include <iostream>
 3. #include "..\sqlserver\SQLSvrPool.h"
 4. 
 5. using namespace std;
 6. 
 7. int main(){
 8. 
 9.         SQLSvrPool* sqlsvrpool = new SQLSvrPool(    "localhost",    //hostnam
10.                                                     "MyDatabase",    //database 
11.                                                     "sa",            //username
12.                                                     "Admin-123",    //password
13.                                                     300000,        //keepalive timeout (milliseconds)
14.                                                     "SELECT top 1 name FROM dbo.sysobjects");    //keepalive statement
15. 
16.         /*Create a pool that will have 3 cached connections and will swell upto a 
17.           total of 5 connections. Returns the number of cached connections or -1 on error
18.          */
19.         if(sqlsvrpool->CreatePool(3, 5)<=0){
20.             cout<<"Error creating database pool\n";
21.             cout<<sqlsvrpool->GetLastSystemError()<<endl;    //If it's asystem error
22.             goto EXAMPLE_END;
23.         }
24. 
25.         /*Dispaly the pool information*/
26.         cout<<(*sqlsvrpool);
27. 
28.         SQLHANDLE* psqlconnectionhandle;
29.         SQLHANDLE  sqlstatementhandle;
30.     
31. 
32.         /*Test the validity of the pool at anytime - not really required*/
33.         if(!sqlsvrpool->IsPoolValid())
34.             goto EXAMPLE_END;
35. 
36.         /*Get a connection from the pool*/
37.         if((psqlconnectionhandle=sqlsvrpool->GetConnectionFromPool())==0){
38.             cout<<"You have reached the maximum amout allowed - 5 in this example\n";
39.             goto EXAMPLE_END;
40.         }
41. 
42.         /*Get a statement handle from the connection*/
43.         if(SQL_SUCCESS!=SQLAllocHandle(SQL_HANDLE_STMT, *psqlconnectionhandle, &sqlstatementhandle)){
44.             sqlsvrpool->ShowSQLError(cout, SQL_HANDLE_DBC, *psqlconnectionhandle);
45.             goto EXAMPLE_END;
46.         }
47. 
48.         /*Execute the query and display the results. Do not Free the database connection handle.
49.           This will be released via the ReleaseConnectionToPool back to the pool.
50.          */
51.         if(SQL_SUCCESS!=SQLExecDirect(sqlstatementhandle, (SQLCHAR*)"select * from testtable", SQL_NTS)){
52.             sqlsvrpool->ShowSQLError(cout, SQL_HANDLE_STMT, sqlstatementhandle);
53.             goto EXAMPLE_END;
54.         }
55.         else{
56.             char name[64];
57.             char address[64];
58.             int id;
59.             while(SQLFetch(sqlstatementhandle)==SQL_SUCCESS){
60.                 SQLGetData(sqlstatementhandle, 1, SQL_C_ULONG, &id, 0, NULL);
61.                 SQLGetData(sqlstatementhandle, 2, SQL_C_CHAR, name, 64, NULL);
62.                 SQLGetData(sqlstatementhandle, 3, SQL_C_CHAR, address, 64, NULL);
63.                 cout<<id<<" "<<name<<" "<<address<<endl;
64.             }
65.         }
66.         SQLFreeHandle(SQL_HANDLE_STMT, sqlstatementhandle );
67. 
68.         /*Dispaly the pool information*/
69.         cout<<(*sqlsvrpool);
70. 
71. 
72.         /*Release the connection back into the pool*/
73.         sqlsvrpool->ReleaseConnectionToPool(psqlconnectionhandle);
74. 
75.         /*Dispaly the pool information*/
76.         cout<<(*sqlsvrpool);
77.         
78.         char c;
79.         cout<<"Enter character to exit\n";
80.         cin>>c;
81. 
82. EXAMPLE_END:
83. 
84.         /*Destroy the database pool*/
85.         if(sqlsvrpool->DestroyPool()>0){
86.             cout<<"There are still some un-released connections in the pool\n";
87.         }
88.         
89.         delete sqlsvrpool;
90. 
91.  return 0;
92. }
93.

Hide line numbers

As you can see the basic steps are..

SQLSvrPool* sqlsvrpool = new SQLSvrPool("hostname", "mydatabase", "username","password", 300000, "SELECT top 1 name FROM dbo.sysobjects");
The keepalive thread will issue the "SELECT top 1 name FROM dbo.sysobjects" statement every 300 seconds across all the inactive connections in the cache to keep them alive.
sqlsvrpool ->CreatePool(3, 5);
This creates a cache with 3 stored connections that will swell to a max of 5 connections when required. If you go above 5 connections without releasing any then you will get a 0 when you next call GetConnectionFromPool().
pconnectionhandle=sqlsvrpool->GetConnectionFromPool()
Do stuff with the connection. Remember not to release or delete this connection as it will be returned to the pool.
sqlsvrpool ->ReleaseConnectionToPool(pconnectionhandle);
sqlsvrpool ->DestroyPool();

Compiling and running the code

Open a command prompt and go to the "examples" direcotory. Compile the files by running nmake. Make sure to substitute your local directories for the ones in the file. There should be the oracle oci include directory and the oraocci10.lib binary for compile.

..\CppDatabasePool\examples>nmake -f makefile.oracle

`
Add these paths to your existing path so that the executable can find the required libraries. Again, don't forget to replace the directories with the ones on your local machine. Basically you need to give the 'bin' diretory and the 'oci\lib\msvc\vc7' directory in your of your Oracle client installation.

..\CppDatabasePool\examples>set path=%path%;D:\oracle\product\10.2.0\client_1\oci\lib\msvc\vc7

..\CppDatabasePool\examples>set path=%path%;D:\oracle\product\10.2.0\client_1\BIN
In the command prompt run 'testoracle.exe.' Make sure the database parameters are your own. ie. username, pasword, tns name etc.

Back to the tutorial trail | Home

Ant Build File and Ant Tutorial

Aim
The aim of this tutorial is to give a quick introduction to Ant. We will look at the basics of a build file and go through some of the options available. Ant buildfile is very similar to nmake/name. Leave any questions or problems as comments and I will endeavour to answer them.

Assumptions
This article assumes that you have a compatible version of java (JDK) and Ant installed.

As an example we will use a simplified version of buildfile that is used to build the Web Services Example Using Axis2 and Tomcat.

 1. <?xml version="1.0" encoding="UTF-8"?>
 2. <project basedir="." default="build" name="Axis2TomcatHelloWorld">
 3.     <property environment="env"/>
 4.     <property name="debuglevel" value="source,lines,vars"/>
 5.     <property name="target" value="1.5"/>
 6.     <property name="source" value="1.5"/>
 7.     <property name="AXIS2_HOME" value="D:/downloads/axis2-1.4.1-bin/axis2-1.4.1"/>
 8.     <property name="HelloWorldAar" value="axis2tomcathelloworld.aar"/>
 9.     <path id="Axis2TomcatHelloWorld1.classpath">
10.         <pathelement location="client_bin"/>
11.         <pathelement location="web_bin"/>
12.         <fileset dir="D:\downloads\axis2-1.4.1-bin\axis2-1.4.1">
13.             <include name="**/*.jar"/>
14.         </fileset>
15.     </path>
16.     <target name="init">
17.         <mkdir dir="web_bin"/>
18.         <copy includeemptydirs="false" todir="web_bin">
19.             <fileset dir="web_src" excludes="**/*.launch, **/*.java"/>
20.         </copy>
21.         <mkdir dir="client_bin"/>
22.         <copy includeemptydirs="false" todir="client_bin">
23.             <fileset dir="client_src" excludes="**/*.launch, **/*.java"/>
24.         </copy>
25.     </target>
26.     <target name="build" depends="build-service, build-client"/>
27.     <target name="clean">
28.         <delete dir="web_bin"/>
29.         <delete dir="client_bin"/>
30.         <delete file="${HelloWorldAar}"/>
31.     </target>
32.     <target name="build-service" depends="init">
33.         <echo message="${ant.project.name}: ${ant.file}"/>
34.         <javac debug="true" debuglevel="${debuglevel}" destdir="web_bin" source="${source}" target="${target}">
35.             <src path="web_src"/>
36.             <classpath refid="Axis2TomcatHelloWorld1.classpath"/>
37.         </javac>
38.     </target>
39.     <target name="build-client" depends="init">
40.         <antcall target="wsdl2java"/>
41.         <javac debug="true" debuglevel="${debuglevel}" destdir="client_bin" source="${source}" target="${target}">
42.             <src path="client_src"/>
43.             <classpath refid="Axis2TomcatHelloWorld1.classpath"/>
44.         </javac>
45.     </target>
46. </project>

Hide line numbers

Windows C++ Producer Consumer Threaded Example

Aim
The aim of this Windows C++ example is to demonstrate multi-threading, signalling, semaphores and mutexes using a simple producer and consumer example in C++ on Windows. This will use WINAPI functions such as CreateThread, WaitForMultipleObjects, CreateSemaphore, CreateMutex, ReleaseMutex and ReleaseSemaphore. Leave any questions or problems as comments and I will endeavour to answer them.

Assumptions
This article assumes that you have VC++ installed and configured. See Problems and Solutions Installing VC++ Express Edition, to install and run vcvars32.bat.

Versions used in this example

Sofware/Component	Image
Windows XP SP2	N/A
Visual Studio Express Editions 2008 VC++	N/A

Links to these files can be found here

Write and compile the application

This application uses an integer queue. The producer adds values to this queue from 1 to 99 by pushing. Each time it adds a value it notifies a thread via a semaphore that there is something to consume. A consumer thread then consumes this value by popping it. Write the application and save is as thread.cpp.

 1. #include <windows.h>
 2. #include <iostream>
 3. #include <queue>
 4. #include <process.h>
 5. 
 6. using namespace std;
 7. 
 8. DWORD WINAPI threadConsumer(void*);
 9. DWORD WINAPI threadProducer(void*);
10. 
11. #define GETMYRAND() (int)(((double)rand()/(double)RAND_MAX)*100)
12. 
13. struct MyData{
14.     queue<int> food;
15.     bool producerfinished;
16.     HANDLE mutex;
17.     HANDLE controlsemaphore;
18. };
19. 
20. int main (){
21.     MyData mydata;
22.     mydata.mutex = CreateMutex(NULL, false, NULL);
23.     mydata.controlsemaphore = CreateSemaphore(NULL,0,2,NULL);
24.     mydata.producerfinished=false;
25. 
26.     HANDLE handles[3];
27. 
28.     handles[0] = CreateThread(0,0,&threadConsumer, (void*)&mydata , 0,0);
29.     handles[1] = CreateThread(0,0,&threadProducer, (void*)&mydata, 0,0);
30.     handles[2] = CreateThread(0,0,&threadConsumer, (void*)&mydata, 0,0);
31. 
32.     WaitForMultipleObjects(3, handles, true, INFINITE); //"Join" trreads
33.     
34.     CloseHandle(mydata.mutex);
35.     CloseHandle(mydata.controlsemaphore);
36. }
37. 
38. DWORD WINAPI threadConsumer(void* lp){
39.     MyData * md = (MyData*)lp;
40.     
41.     while(!md->producerfinished){
42. 
43.         Sleep(GETMYRAND());
44. 
45.         WaitForSingleObject(md->controlsemaphore,INFINITE); //Wait for producer;
46. 
47.         WaitForSingleObject(md->mutex, INFINITE); //Mutext Lock
48.         while(md->food.size()>0){
49.             char tmps[32];
50.             sprintf(tmps, "\t%d\t(%d)\n",md->food.front(), GetCurrentThreadId());
51.             cout<<tmps;
52.             md->food.pop();
53.         }
54.         ReleaseMutex(md->mutex); //Mutex Unlock
55. 
56.     }
57.     return 0;
58. }
59. 
60. DWORD WINAPI threadProducer(void* lp){
61.     MyData * md = (MyData*)lp;
62. 
63.     for(int i =0 ; i < 100; i++){
64.         char tmps[32];
65.         sprintf(tmps, "%d\t\t%d\n",i,GetCurrentThreadId());
66.         cout<<tmps;
67.     
68.         Sleep(GETMYRAND());
69. 
70.         WaitForSingleObject(md->mutex, INFINITE); //Mutex Lock
71.         md->food.push(i);
72.         if(md->food.size()>0)
73.             ReleaseSemaphore(md->controlsemaphore,1, NULL);
74.         ReleaseMutex(md->mutex); //Mutex Unlock
75.     }
76.     md->producerfinished=true;
77.     ReleaseSemaphore(md->controlsemaphore,2, NULL); //Release both threads
78.     return 0;
79. }

Hide line numbers

Line11: Randomly generate a sleep time between 0 and 100 milliseconds.

Lines 28-30: Start all the threads giving the same MyData variable to them so the semaphore and the mutex is shared between all of them.

Line 45 : Both consumer threads wait on the semaphore.

Line 73: The producer thread triggers the semaphore allowing one waiting thread to proceed.

Lines 47 & 54: Lock and unlock mutexes allowing only a single consumer thread to run this section of code. This is like a critical section or a 'synchronized' section java.

Line 32: Wait for all the threads to finish processing. Like the join command in Linux.

Open a prompt to the working directory and compile the code using the windows cl.exe compiler. You just need to include the pdh.lib.

..workspace\WinapiThreadExample>cl thread.cpp

Now run the program in the command prompt. You should see something similar to what is shown below.


80              2724
        79      (2996)
81              2724
        80      (1108)
        81      (2996)
82              2724
83              2724
        82      (1108)
84              2724
        83      (2996)
85              2724
        84      (1108)
86              2724

The first column is the values produced by the producer, the second column is the value consumed by the consumer and the third column is the threadid of the consumers (in brackets) and producer. Note that there are 2 consumers in this example, hence the two consumer thread IDs 2996 and 1108 in this example

C++ Home	Java/javaEE Home	Hints & Tips
Create Widgets	Javascript/Perl/ Web2 Home