Changelog.txt

* 04/07/12 - v.2.0.0xsxsx
   - Currently distributed communication pattern implemented are: 
     Unicast, Broadcast, Scatter, Gather_All, FromAny, OnDemand
   - first distributed implementation based on ZeroMQ message library
   - dnode class added, minor changes to node class
* not-released - v.1.2.0
   - added a new implementation of the unbounded MPMC queue
   - added an implementation of the bounded Multi-Producer/Multi-Consumer 
     queue by Dmitry Vyukov (www.1024cores.net)
   - implemented all_gather
   - moved from int to unsigned long for embedded atomic operations
   - performance improvements in the ff_allocator for the test one to many
   - Micheal Scott 2-locks MPMC queue implementation (two methods mp_push 
     and mp_pop have been added to the dSPSC queue)
* 22/04/11 - v.1.1.1
   - accelerator support implemented for the pipeline
   - ff_send_out improvment: more control over the number of retries, return
     value added
   - barrier protocol re-worked, now is possible to create and to start a 
     pipeline or a farm (or an accelerator) from inside a pipeline's node or 
     a farm's worker/emitter/collector 
   - added ffwTime which considers only the time spent in the svc methods
   - changed the value of FF_GO_ON
   - fixed SWSR_Ptr_Buffer length method (unsigned long --> long)
   - fixed a bug which prevents to have a farm whose workers are D&C skeleton
     (see test_multi_masterworker test)
   - added the possibility to define workers' input queue size for the 
     on-demand scheduling policy (default value is 1 slot)
* 22/04/11 - v.1.1.0
   - MSqueue reworked, now it is not in the experimental state.
     The queue is working also on Windows OS and OSX. Removed DCAS and used CAS.
   - removed some performance problems on Windows OS. 
   - slightly modified MPMC interface and usage
   - added Quicksort algo implementation which uses MPMC queue 
     (both in blocking and non-blocking version)
   - fixed some compilation problem under OSx
* 01/04/11 - v.1.0.9
   - fixes return type error in node.hpp ubuffer.hpp and buffer.hpp
   - added the method get_channel_id() in the emitter and collector class
     the method returns the id of the worker thread from which an input task
     is coming from
   - added the fibonacci and the quicksort examples
   - added basic interface for pinning threads to CPUs
   - added multipush and mpush to SWSR buffer (experimental code)
   - added mpush to uSWSR queue (experimental code)
   - more tests
   - improved cmake test
   - improved performance for fine grain D&C computation (farm+feedback)
     by changing the losetime_in method in the lb class.
   - added Multi-Producers/Multi-Consumers queue (MSqueue by Michael and Scott) 
     (EXPERIMENTAL code )
   - added abstraction_dcas from liblfds (www.liblfds.org), no powerpc support yet!
   - fixed some bugs in the allocator
   - posix_memalign implemented in the allocator
   - ticket-spinlock from Linux kernel (EXPERIMENTAL code only for Linux)
   - added two lock-based methods in uSWSR queue (mp_push and mc_pop)
   - fixed missing files in CMakeLists.txt
   - fixed deadlock situation when pipeline with feedback channel is created (torus pattern)
   - implemented a Deferred Reclamation strategy based on number of free in the allocator
   - added the ff_queue implementation of a SPSC queue by Dmitry Vyukov (www.1024cores.net)
   - first porting on Windows OS
* 01/09/10 - v.1.0.0
   - New release
   - some bugs fixed
   - unbounded SWSR queue improved (removed all locks)
   - added dynqueue (dynamic list-based queue)
   - all .hpp files moved into the ff directory
   - more tests added
   - cmake compilation support (thanks to Fedor Sakharov)
   - improved the accelerator sturcture (added FF_EOS_NOFREEZE tag)
   - added the management of second level streams
   - added the 'stop' method in the farm and pipeline skeleton
   - fixed ffStats method when run_then_freeze is called multiple times
   - one memory leak removed
   - removed some warnings related to strict-aliasing 
* 22/03/10 - v.1.0.0rc2
   - Moved to LGPLv3 license. 
   - More tests and applications (including pbzip2). 
   - FastFlow Allocator improved. 
   - Fixed some minor bugs. 
   - Added the broadcast_task method.
* 03/02/10 - v.1.0.0rc1 
   - Minor API revision. 
   - Improved FastFlow accelerator support. 
   - More tests and applications,(Nokia QT Mandelbrot and NQueens)
   - Simple execution trace support.
   - Allocator improved. 
* 16/11/09 - v.0.9.7
   - Major API revision: patterns are no longer object factories but objects. 
   - Support for arbitrary nesting of pipe, farm, and loop at high-level layer.
   - First version of the Divide&Conquer pattern (no examples yet).
   - First version of the FastFlow's Accelerator.
* 19/10/09 - v.0.6.1
   - FastFlow-swps3 Smith-Waterman application added.
* 15/10/09 - v.0.5.0
   - First release. 
   - Support for farm skeleton/pattern.