• A producer computes and outputs a stream of results.
• A consumer inputs and analyzes a stream of values.
• A pipeline can be seen as a sequence of producers-consumers. Each process in a pipeline is a filter that consumes the output of the previous process and produces input for the next process. Example: Unix pipelines

ls -Al /usr/include/ | more

General model

• Suppose that the data are in a stack of limited size (an array). We will suppose that the order in which the data are processed doesn't matter.
• There is an index, counter, that tells us which is the size of the current data.
• The data is supposed to be accessible to all of the processes - shared memory.
• For message-passing, the data will be manipulated by the master, which can be implemented as a server.

Producer()
{
  while (true) {
    item = Produce_item();
    while (!Deposit(item)) ;
  }
}

Consumer()
{
  while (true) {
    while (!Fetch(item));
    Consume(item);
  }
}

int Deposit(some_type item)
{
  if (Data_is_full())
    return 0;
  else {
    data[counter++] = item;
    return 1;
  }
}

int Fetch(some_type &item)
{
  if (Data_is_empty())
    return 0;
  else {
    item = data[--counter];
    return 1;
  }
}

int Data_is_full()
{ return (counter == max); }

int Data_is_empty()
{ return (counter == 0); }

The code of the program using pthreads:

producer_consumer.cc
producer_consumer.h