Re: [Boost-users] Trying to get Boost.Interprocess going on OS X 10.5.1

From: Stefan Arentz (stefan.arentz_at_[hidden])
Date: 2007-11-25 07:52:46

On 11/25/07, Ion Gaztañaga <igaztanaga_at_[hidden]> wrote:
> Stefan Arentz escribió:
> > I checked out boost trunk from svn and tried to compile a simple
> > boost.interprocess example. Unfortunately it fails at a rather
> > unexpected place:
> >
> > % c++ doc_message_queueA.cpp -Itrunk -o doc_message_queueA
> > % ./doc_message_queueA
> > Assertion failed: (res == 0), function lock, file
> > trunk/boost/interprocess/sync/posix/interprocess_mutex.hpp, line 116.
> > zsh: abort      ./a
> >
> > The error code is 22, EINVAL.
> >
> > I'm on OS X 10.5.1, maybe that is too new, or not yet tested?
>
> It is not tested but your problem and another problem reported a couple
> of weeks ago are pretty strange. If Leopard now offers process-shared
> posix mutexes and native shared memory, in theory Interprocess code
> should work without problems, just like other Unixes, but I don't know
> what's happening.
>
> I don't have access to the OS so I'll need some help to fix these
> issues. Meanwhile, try to undef BOOST_INTERPROCESS_POSIX_PROCESS_SHARED
> define <boost/interprocess/detail/workaround.hpp> line 21. This will
> avoid native process-shared mutexes and conditions and Interprocess will
> work as in Mac Os 10.4

Thanks Ion. That did the trick. I'll check if Leopard actually has
process-shared mutexes.

I have another problem now. I am trying to create a simple
cross-process consumer/producer example based on the
doc_message_queue*.cpp examples.

My producer looks like this:

      message_queue mq(create_only, "message_queue", 3600, sizeof(int));

      for(int i = 0; i < 3600; ++i){
         std::cout << "Publishing " << i << std::endl;
         mq.send(&i, sizeof(i), 0);
         sleep(1);
      }

And the consumer like this:

      message_queue mq(open_only ,"message_queue");

      unsigned int priority;
      std::size_t recvd_size;

      for(int i = 0; i < 10; ++i){
         int number;
         mq.receive(&number, sizeof(number), recvd_size, priority);
         std::cout << "Consumed " << number << std::endl;
      }

When I start the producer it starts filling the queue. When I run the
consumer multiple times it first quickly grabs 10 numbers in the queue
until it is empty and then i see the rest appear every second. So far
so good.

The problem is that when I kill or Control-C the consumer that the
producer stops. I assume that this is because some lock is held by the
consumer and the producer is waiting for it.

Is there a way to release these locks when the consumer is killed? Or
a way to detect this, so that I can release them manually or so?

 S.