Previous | Contents | Index |
The maximum size of the queue does not affect the amount of memory occupied by the queue when the queue is empty.
qhead *cut()
Splits a queue into two queues. One queue has a new qhead object, which the return value points to, and the original qtail object; it contains the objects from the original queue. The other queue has the original qhead object and a new qtail object; this queue is empty. You can use this function to insert a filter into an existing queue without changing the queue's appearance to functions that access the ends of the queue, and without halting the flow through the queue of objects.object *get()
Returns a pointer to the object at the head of the queue when the queue is not empty. The object is removed from the queue. If the queue is empty, behavior depends on the mode of the qhead object. In WMODE, a task that executes qhead::get() on an empty queue suspends until that queue is not empty. In EMODE, executing qhead::get() on an empty queue causes a run-time error. In WMODE, executing qhead::get() on an empty queue returns the NULL pointer instead of a pointer to an object.virtual objtype o_type()
Returns object::QHEAD.int pending()
Specifies that get operations on a queue must wait until an object is put in the queue. It returns a nonzero value if the queue attached to a qhead object is empty; otherwise, it returns 0.void print (int verbosity, int internal_use = 0)
Prints a qhead object on cout. The verbosity argument specifies the information to be printed. Do not supply a value for the internal_use parameter.int putback(object *new_queue_element)
Inserts at the head of the queue the object that the new_queue_element argument points to, and returns a value of 1 on success. This lets the qhead object operate as a stack (hence, the name putback). Space must be available in the queue for it to succeed. Calling qhead::putback() for a full queue causes a run-time error in both EMODE and WMODE and returns NULL in ZMODE.int rdcount()
Returns the current number of objects in the queue attached to a qhead object.int rdmax()
Returns the maximum length of the queue.qmodetype rdmode()
Returns the current mode of a qhead object, which can be EMODE, WMODE, or ZMODE.void setmode(qmodetype modetype)
Sets the mode of a qhead object to modetype, which can be EMODE, WMODE, or ZMODE.void setmax(int size)
Sets size as the maximum length of the queue attached to a qhead object. You can set size to a number less than the current number of objects of the object class, but that means you cannot put any more objects of the object class on the queue until the length of the queue has been reduced below the limit you set.void splice(qtail *delete_tail)
Forms a single queue by appending a queue attached to a qhead object onto the queue referenced in the argument. Typically, this reverses the action of a previous qhead::cut() function. The extra qhead and qtail objects are deleted. Waiting tasks resume execution if merging the two creates a nonempty queue (if the task was trying to get) or an empty queue (if the task was trying to put).qtail *tail()
Creates a qtail object for the queue attached to a qhead object (if none exists) and returns a pointer to the new qtail object.
Abstraction for the tail of a list of items in a first-in, first-out singly linked list.
#include <task.hxx>Alternative Header
#include <task.h>
class qtail: public object { friend class qhead; public: qtail(qmodetype modetype = WMODE, int size = 10000); ~qtail(); qtail *cut(); qhead *head(); int put(object *new_queue_element); int rdspace(); int rdmax(); qmodetype rdmode(); void setmode(qmodetype modetype); void setmax(int size); void splice(qhead *delete_head); int pending(); void print(int verbosity, int internal_use = 0); objtype o_type(); };
This class provides facilities for putting objects into a queue. A queue is a data structure with an associated list of objects of the object class, or a class derived from the object class in first-in, first-out order. All access to a queue is through either the attached qhead or qtail object. You create a queue by creating either a qhead or a qtail object. The other end of the queue is created automatically. You can then obtain a pointer to the head with the qtail::head function.Objects have definitions for when they are ready and pending (not ready). The qtail objects are ready when the queue is not full and pending when the queue is full.
When a run-time error occurs, the appropriate error code from the following table is passed to the object::task_error() function:
Value Error Description e_putfull Cannot put an object into a full queue e_putobj Cannot put an object into queue if the object is on another queue e_qdel Cannot delete a queue that has an object in the queue e_store Cannot allocate more memory
qtail(qmodetype modetype = WMODE, int size = 10000)
Constructs a qtail object. The modetype argument specifies the mode (set by the constructor) that controls what happens when an object of the qtail class is pending. The choices are WMODE (wait mode), EMODE (error mode), or ZMODE (0 mode); WMODE is the default. (See the put() function for more information.) The size argument specifies the maximum length of the queue attached to a qhead object; the default is 10,000.The maximum size of the queue does not affect the amount of memory occupied by the queue when the queue is empty.
~qtail()
Deletes a qtail object.
qtail *cut()
Splits a queue into two queues. One queue has a new qtail object (to which the return value points) and the original qhead object; it contains the objects from the original queue. The other queue has the original qtail object and a new qhead object; this queue is empty. You can use this function to insert a filter into an existing queue, without changing the queue's appearance to functions that access the ends of the queue, and without halting the flow through the queue of objects.qhead *head()
Creates a qhead object for the queue attached to a qtail object (if none exists) and returns a pointer to the new qhead object.virtual objtype o_type()
Returns object::QTAIL.int pending()
Specifies that get operations on a queue must wait until an object is put in the queue. It returns a nonzero value if the queue is empty; otherwise, it returns 0.virtual void print(int verbosity, int internal_use = 0)
Prints a qtail object on cout. The verbosity argument specifies the information to be printed. Do not supply a value for the internal_use parameter.int put(object *new_queue_element)
Adds the object denoted by the new_queue_element argument to the tail of the queue attached to a qtail object; returns a value of 1 on success. If the queue is full, the behavior depends on the mode of the qtail object. In WMODE, an object of class task that executes qhead::put() on a full queue suspends until that queue is not full. Calling qhead::put() for a full queue causes a run-time error in EMODE and returns NULL in ZMODE.int rdspace()
Returns the number of object objects that can be inserted into the queue before it becomes full.int rdmax()
Returns the maximum length of the queue.qmodetype rdmode()
Returns the current mode of a qtail object, which can be EMODE, WMODE, or ZMODE.void setmode(qmodetype modetype)
Sets the mode of a qtail object to modetype, which can be EMODE, WMODE, or ZMODE.void setmax(int size)
Sets size as the maximum length of the queue. You can set size to a number less than the current number of objects of the object class, but that means you cannot put any more objects of the object class on the queue until the length of the queue has been reduced below the limit you set.void splice(qhead *delete_head)
Forms a single queue by appending a queue attached to a qtail onto the queue referenced in the argument. Typically, this reverses the action of a previous qtail::cut(). The extra qhead and qtail objects are deleted. Waiting tasks resume execution if merging the two queues creates a nonempty queue (if the task was trying to get) or an empty queue (if the task was trying to put).
Objects of the randint class generate uniformly distributed random numbers.
#include <task.hxx>Alternative Header
#include <task.h>
class randint { public: randint(long seed=0); int draw(); float fdraw(); void seed(long seed); };
Objects of this class generate uniformly distributed random numbers. Each random-number generator object produces a sequence that is independent of other random-number generator objects.
randint(long seed)
Constructs an object of the randint class. The seed argument is used as the seed and is optional. Different seeds produce different sequences of generated numbers; not all seeds produce useful sequences.
float fdraw()
Returns the next random number generated by the object. The number is a floating-point value in the range 0 to 1.int draw()
Returns the next random number generated by the object. The number is an integer value in the range from 0 to RAND_MAX, which is defined in the ANSI C header, stdlib.h.void seed(long seed)
Reinitializes the object with the seed seed.
extern "C" { #include <stdlib.h> } #include <task.hxx> #include <iostream.hxx> main() { randint gen; int i=0; float sum; for (i=0; i<1000; i++) sum += gen.fdraw(); cout<<"Average is " << sum/1000. << "\n"; return EXIT_SUCCESS; } |
This example prints the average of 1000 floating-point random numbers.
Responsible for scheduling and for the functionality common to task and timer objects.
#include <task.hxx>Alternative Header
#include <task.h>
class sched: public object { public: enum statetype { IDLE = 1, RUNNING = 2, TERMINATED = 4 }; protected: sched(); public: static task *clock_task; static PFV exit_fct; void cancel(int result); int dont_wait(); sched *get_priority_sched(); int keep_waiting(); statetype rdstate(); long rdtime(); int result(); int pending(); virtual void print(int verbosity, int internal_use = 0); virtual void setwho(object *alerter); static long get_clock(); static sched *get_run_chain(); static int get_exit_status(); static void set_exit_status(int); static void setclock(long); }; #ifdef CXXL_DEFINE_CLOCK #define clock (sched::get_clock()) #endif #define run_chain (sched::get_run_chain())
This class provides facilities for checking on the state of a task, manipulating the simulated clock, canceling a task, and checking on the result of a task.You can create instances of classes derived from the sched class, but you cannot create instances of the sched class itself.
When a run-time error occurs, the appropriate error code from the following table is passed to the object::task_error() function:
Value Error Description e_clockidle Cannot advance the clock when the clock_task is running or terminated e_negtime Cannot delay a negative amount of time e_resobj Cannot resume a task or timer if it is already on another queue e_resrun Cannot resume a running task e_resterm Cannot resume a terminated task e_schobj Cannot use class sched other than as a base class e_schtime Cannot execute something at a time that has already passed e_setclock Cannot set the clock after it has advanced past 0
static task *clock_task
Points to the task clock if one exists.static PFV exit_fct
Points to the exit function if one exists.
sched()
Constructs a sched object initialized to the IDLE state and delay 0.
void cancel(int result)
Puts an object into the TERMINATED state without suspending the caller (that is, without invoking the scheduler); sets the result of the object to result.int dont_wait()
Returns the number of calls to keep_waiting(), minus the number of calls to the dont_wait() function, excluding the current call. The return value of this function should equal the number of objects of the object class waiting for external events before the current dont_wait() call.long get_clock()
Returns the value of the clock in simulated time units.int get_exit_status()
Returns the exit status of the task program. When a task program terminates successfully (without calling task_error), the program calls exit(i) where i is the value passed by the last caller of sched::set_exit_status().sched *get_priority_sched()
Returns a pointer to a system task's interrupt_alerter if the system gets an awaited signal. If no interrupt occurs, this function returns 0.sched *get_run_chain()
Returns a pointer to the run chain, the linked list of ready objects belonging to classes derived from the sched class (task and timer objects).int keep_waiting()
Keeps the scheduler from exiting when no tasks exist that can be run (an external event could enable an IDLE task to be run). This function should be called when the user program creates an object that waits for an external event. Afterward, when such an object destructs, a call should go to the dont_wait() function. The keep_waiting() function returns the number of calls (not counting the current call) minus the number of calls to the dont_wait() function.int pending()
Returns 0 if the object is in the TERMINATED state; otherwise, it returns a nonzero value.virtual void print(int verbosity, int internal_use = 0)
Prints a sched object on cout. The verbosity argument specifies the information to be printed. Do not supply a value for the internal_use parameter.statetype rdstate()
Returns the state of the object: RUNNING, IDLE, or TERMINATED.long rdtime()
Returns the simulated clock time at which to run the object.int result()
Returns the result of a sched object (as set by the task::resultis(), task::cancel(), or sched::cancel() function). If the object is not yet TERMINATED, the calling task suspends and waits for the object to terminate. A task calling result() for itself causes a run-time error.void setclock(long new_clock)
Initializes the simulated clock to a time specified by the new_clock argument. You can use this function once before the simulated clock has advanced without causing a run-time error. To advance the clock after the initial setting, call the task::delay function.void set_exit_status(int new_exit_status)
Sets the exit status of the task program. When a task program terminates successfully (without calling task_error), the program calls exit(i), where i is the value passed by the last caller of sched::set_exit_status().virtual void setwho(object *alerter)
Records which object alerted the object. The alerter argument should represent a pointer to the object that caused the task package to alert the sched.
The DEC C++ Class Library supplies the following macros for compatibility with older C++ class library implementations:clock
Calls sched::get_clock(). For this macro to be defined, you must define CXXL_DEFINE_CLOCK on the command line when invoking the compiler, or in your source code before including the task package header.run_chain
Calls sched::get_run_chain().
Serves as the basis for coroutines.
#include <task.hxx>Alternative Header
#include <task.h>
class task: public sched { public: enum modetype { DEDICATED = 1, SHARED = 2 }; protected: task(char *name = (char *)NULL, modetype mode = DEFAULT_MODE, int stacksize = 0); public: task *t_next; char *t_name; ~task(); void cancel(int); void delay(long); long preempt(); void resultis(int); void setwho(object *); void sleep(object *object_waiting_for = (object *)NULL); void wait(object *); int waitlist(object * ...); int waitvec(object **); object *who_alerted_me(); virtual void print(int verbosity, int internal_use = 0); virtual objtype o_type(); static task *get_task_chain(); };
This class is used only as a base class; all coroutine classes are derived from it. All work for an object of a given coroutine type occurs within the constructor for that type. The coroutine class must be exactly one level of derivation from the task class. When the object is created, the constructor takes control and runs until halted by one of the following functions: wait(), waitlist(), waitvec(), sleep(), or resultis().When a task executes a blocking function on an object that is ready, the operation succeeds immediately and the task continues running; if the object is pending, the task waits. Control then returns to the scheduler, which selects the next task from the ready list or run chain. When a pending object becomes ready, the system puts any task waiting for that object back on the run chain.
A task can be in one of the following states:
RUNNING Running or ready to run IDLE Waiting for a pending object TERMINATED Completed; not able to resume running (but you can retrieve the result)
When a run-time error occurs, the appropriate error code from the following table is passed to the object::task_error() function:
Value Error Description e_result Cannot call result() on thistask e_stack Cannot extend stack e_store Cannot allocate more memory e_taskdel Cannot delete a task that is idle or running e_taskmode Cannot create a task with a mode other than dedicated or shared e_tasknameoverrun Internal error: data overrun when building default task name e_taskpre Cannot preempt a task that is idle or terminated e_wait Cannot call wait() on thistask
task *t_next
Points to the text task on the chain of all task objects; it is equal to NULL if there are no more tasks.char *t_name
Points to the null-terminated task name passed to the constructor. If no name was passed to the constructor, then the constructor creates a unique name (and t_name points to it). If the constructor created the name, then the destructor deletes the name.
task(char *name = (char *)NULL, modetype mode = DEFAULT_MODE,
Constructs a task object. All three arguments are optional and have default values. If you supply a character pointer, name is used as the task object's name. The argument mode must be DEDICATED or SHARED (or omitted) but only DEDICATED is implemented; thus, the mode argument has no effect. The argument stacksize specifies the minimum size of the task object's stack. By default, the stack size is the same as the default for the underlying thread system.
int stacksize = 0)
Previous Next Contents Index