/* * Copyright (C) 2007-2011 Xagasoft, All rights reserved. * * This file is part of the libbu++ library and is released under the * terms of the license contained in the file LICENSE. */ #ifndef BU_THREAD_H #define BU_THREAD_H #include namespace Bu { class ThreadId { friend class Thread; private: ThreadId( pthread_t tId ); public: ThreadId(); bool operator==( const ThreadId &rhs ); bool operator!=( const ThreadId &rhs ); private: pthread_t tId; }; /** * Simple thread class. This wraps the basic pthread (posix threads) * system in an object oriented sort of way. It allows you to create a * class with standard member variables and callable functions that can be * run in it's own thread, one per class instance. *@ingroup Threading */ class Thread { public: /** * Construct an Thread thread. */ Thread(); /** * Destroy an Thread thread. */ virtual ~Thread(); static ThreadId currentThread(); /** * Begin thread execution. This will call the overridden run function, * which will simply execute in it's own thread until the function * exits, the thread is killed, or the thread is cancelled (optionally). * The thread started in this manner has access to all of it's class * variables, but be sure to protect possible multiple-access with * ThreadMutex objects. * @returns True if starting the thread was successful. False if * something went wrong and the thread has not started. */ bool start(); /** * Forcibly kill a thread. This is not generally considered a good * thing to do, but in those rare cases you need it, it's invaluable. * The problem with stopping (or killing) a thread is that it stops it * the moment you call stop, no matter what it's doing. The object * oriented approach to this will help clean up any class variables * that were used, but anything not managed as a member variable will * probably create a memory leak type of situation. Instead of stop, * consider using cancel, which can be handled by the running thread in * a graceful manner. *@returns True if the thread was stopped, false otherwise. When this * function returns the thread may not have stopped, to ensure that the * thread has really stopped, call join. */ bool stop(); /** * Join the thread in action. This function performs what is commonly * called a thread join. That is that it effectively makes the calling * thread an the Thread thread contained in the called object one in the * same, and pauses the calling thread until the called thread exits. * That is, when called from, say, your main(), mythread.join() will * not return until the thread mythread has exited. This is very handy * at the end of programs to ensure all of your data was cleaned up. *@returns True if the thread was joined, false if the thread couldn't * be joined, usually because it isn't running to begin with. */ bool join(); private: pthread_t ptHandle; /**< Internal handle to the posix thread. */ int nHandle; /**< Numeric handle to the posix thread. */ protected: /** * The workhorse of the Thread class. This is the function that will run * in the thread, when this function exits the thread dies and is * cleaned up by the system. Make sure to read up on ThreadMutex, * ThreadCondition, and cancel to see how to control and protect * everything you do in a safe way within this function. *@returns I'm not sure right now, but this is the posix standard form. */ virtual void run()=0; /** * This is the hidden-heard of the thread system. While run is what the * user gets to override, and everything said about it is true, this is * the function that actually makes up the thread, it simply calls the * run member function in an OO-friendly way. This is what allows us to * use member variables from within the thread itself. *@param pThread Should always be this. *@returns This is specified by posix, I'm not sure yet. */ static void *threadRunner( void *pThread ); void yield(); }; } #endif