0.19 20 September 2010

Checking tool chain with perl.5.13.5.

0.18 3 October 2007

        Hopefully fixed test-suite for none thread enabled Perls.  Updated
        to use new VERSION feature of Devel::Required.

0.17 7 January 2004

        Fixed some documentation inconsistencies.  Added documentation on what
        is different from Thread::Queue::Any.

        Made default optimization dependent on the version of Perl used.
        Added test for it.

        Removed mention of Storable: replaced it with Thread::Serialize.
        Added Thread::Serialize as a dependency, just for documentation sake
        really.

0.16 28 December 2003

Added automatic required modules update using Devel::Required.

Added dependency to load.pm.

0.15 9 November 2003

        A problem in (threaded) 5.8.2 made Thread::Conveyor segfault in the
        test-suite.  The problem was fortunately easy to circumvent: just don't
        use subroutines consisting of {} as an object method (in the case of
        Thread::Conveyor, the shutdown() method).

0.14 11 August 2003

        Cleaned up Makefile.PL and updated copyright info.  Verified everything
        works with 5.8.1.  Updated dependency of Thread::Tie.

0.13 30 September 2002

        Use "load.pm" instead of AutoLoader.  Changed dependency on Thread::Tie
        to version 0.08 (which also uses load.pm).

        Added check for availability of methods again: load.pm does support
        that, whereas AutoLoader does not.

0.12 27 September 2002

Ask found another problem in T::C::Tied::clean_dontwait. Fixed.

        Also left warnings enabled in the test-suite last time, which caused
        warnings in the make test.  Changed code so that the warnings are not
        emitted anymore.

0.11 26 September 2002

Fixed some warnings in T::C::Throttled, spotted by Ask Bjoern Hansen.

        Removed "our" from $VERSION and @ISA, should shave off some bytes in
        memory usage, as found from testing with Benchmark::Thread::Size.

0.10 1 September 2002

        Replaced dependency on Thread::Serialize by dependency on Thread::Tie
        as the default, memory optimized implementation, now uses Thread::Tie.

        Changed Thread::Conveyor so that it uses T::C::Tied for unthrottled
        belts when optimizing for memory.

        Changed T::C::Throttled so that it makes its own decision on which
        belt implementation to be used, either T::C::Array (optimized for
        CPU) or T::C::Tied (optimized for memory).

        Added "semaphore" object method to T::C::Array for obtaining the
        lock() semaphore, needed by T::C::Throttled.

        Removed T::C::Thread from the distribution.  This shaves about 1200
        bytes off of the distribution package.

        Replaced Thread::Conveyor::Thread by Thread::Conveyor::Tied, which
        uses Thread::Tie as its shared array implementation.  This has the
        side-effect to being almost twice as slow but saving 18% of
        memory (at least in the test-suite).  So there is now a true CPU
        versus memory trade-off.  Main difference in CPU is caused by the
        fact that whole array fetches aren't possible using the tie()
        interface (this is a loop with FETCH()), wherease T::C::Thread was
        optimized to deal with that case).  Main difference in memory is
        caused by the fact that only one thread (the Thread::Tie default
        thread) is used for all belts, whereas before each belt had its own
        thread.  So the savings in memory could in real world situations, be
        quite a lot more than the 18% seen in the test-suite.

0.09 30 August 2002

        Changed all modules to use AutoLoader to defer loading of necessary
        subroutines to when they are actually needed.  This should save
        memory and CPU for larger programs, or with large numbers of threads.
        The test-suite only marginally takes more memory at the same CPU
        usage: overhead of compiling is levelled out with the overhead of
        cloning pre-compiled routines.

0.08 20 August 2002

        Dave Mitchell showed me the error of my ways with regards to client
        server thread programming.  Fixed the handler optimized for memory so
        that no yield()s are necessary anymore.  The whole thing now gets
        through the test suite more than 25% faster for that part.
        Scalability to many more threads should now be a lot better too.

0.07 14 August 2002

        Removed (possibly premature) optimization from T::C::Thread
        _handler: it now doesn't zap the local threadspace anymore.

0.06 13 August 2002

        Handed over freezing and thawing operations to Thread::Serialize.
        Added dependency to Thread::Serialize in Makefile.PL.

0.05 6 August 2002

        Fixed "onbelt" in T::C::Thread so that it won't hang if the belt is
        shut down.  Needed for Thread::Pool and in general useful.

        5 August 2002
        Added index functionality to "peek" and "peek_dontwait" to all
        sub-classes and added documentation for it.  Needed for Thread::Pool.

        Finally got the T::C::Thread implementation getting through the
        test-suite reliably.  Problem was caused by my misconception that
        cond_wait would always regain the lock() after being signalled.  It
        does not.  This should have been in big red letters in the
        threads::shared pod.

        Added methods "shutdown", "thread" and "tid" plus documentation.

        3 August 2002
        Adapted test-suite to use many more combinations of possible
        settings.

        2 August 2002
        Added new "optimize" field parameter to allow you to select an
        implementation optimized for speed (T::C::Array and T::C::Throttled)
        or for memory (T::C::Thread).

        Added new implementation of Thread::Conveyor based on using a
        seperate thread for keeping the belt: Thread::Conveyor::Thread.

        Move basic parts of Thread::Conveyor to Thread::Conveyor::Array.
        Which should allow for other implementations of the conveyor
        metaphor.

0.04 1 August 2002

        Added internal method "_belt" to get at the actual belt, for both
        the normal and the throttled belt.

0.03 31 July 2002

        Added internal methods "freeze" and "thaw" so that we can move
        to a different serialization scheme or other optimizations without
        having to change all the sub-classes as well.  First user is
        Thread::Conveyor::Monitored.

        Added "maxjobs" and "minjobs" method in Thread::Conveyor that will
        die when they're called (throttling only takes place in
        Thread::Conveyor::Throttled).

        Fixed problem that would cause throttling to not be switched off
        when the "maxjobs" method was specified with an undef value.

        Added methods "clean" and "clean_dontwait" for cleaning the belt
        to both Thread::Conveyor and Thread::Conveyor::Throttled.  Also
        added documentation for it.

        Added internal method "_clean" for obtaining frozen contents of
        the belt to both Thread::Conveyor and Thread::Conveyor::Throttled.
        This was needed for Thread::Conveyor::Monitored.

0.02 30 July 2002

        Adapted Thread::Conveyor to default to a throttled conveyor belt
        automatically if no specific parameters specified.  Now allows for
        specification of throttling settings in a parameter hash reference.
        If unthrottling is specifically specified, returns to the old
        unthrottled behaviour (where the object is just an array reference),
        thus keeping the old execution speed.

        Added helper class Thread::Conveyor::Throttled to allow for throttled
        conveyor belts that block if there are too many boxes on the belt.

        Fixed some documentation nits.

0.01 25 July 2002

        First version of Thread::Conveyor, which started life as
        Thread::Queue::Any.