[Vortex] [ANN] Vortex Library 1.1.6 'Break on Through' is ready!

Francis Brosnan Blazquez francis at aspl.es
Tue Oct 26 19:59:13 CEST 2010


      ###################################################
      ##              Vortex Library 1.1.6             ##
      ##               "Break on Through"              ##
      ###################################################

Release note
~~~~~~~~~~~~

  Advanced Software Production Line is proud to announce Vortex
  Library 1.1.6 "Break on Through" release availability: an open
  source "BEEP Core" protocol implementation written in C, mapped onto
  TCP/IP.

  Vortex Library is a stable and robust application protocol
  development toolkit, with active support for Microsoft Windows and
  GNU/Linux platforms.

  Check out the following links to get ready with Vortex Library:

    - http://www.aspl.es/vortex
      [Vortex Library homepage]

    - http://www.aspl.es/vortex/py-vortex/html
      [PyVortex homepage]

    - http://www.aspl.es/turbulence
      [Turbulence BEEP server homepage]
 
    - http://www.sf.net/projects/vortexlibrary
      [Sources and Win32 binaries]

    - http://www.beepcore.org
      [BEEP Core community web site]

    - http://www.aspl.es/xml
      [Axl Library, XML 1.0 implementation]
    
This release in short
~~~~~~~~~~~~~~~~~~~~~

  This release fixes several issues with the vortex ALIVE
  implementation and also adds python interface to it. ALIVE
  implementation was updated to also detect connection activity to
  skip doing checks in that case.

  Many updates to vortex engine to allow calling to terminate the
  vortex context (VortexCtx) from inside its handlers (frame received
  or connection close). Making vortex connection close notification to
  not block critical threads (vortex sequencer and vortex reader).

  Several updates to PyVortex API.

Change notifications
~~~~~~~~~~~~~~~~~~~~
  None

Thanks to..
~~~~~~~~~~~

  - Benoit Amiaux (reported by fixed double deallocation problem at
    profiles list mutex, typo warning at __vortex_connection_new
    returning axl_false instead of NULL, wrong memory access at
    vortex_tls_get_digest_sized win32 implementation, a double frame
    deallocation when frame received handler failed)

  - Balazs Tirpak (reported and provided initial useful patches to fix
    wrong reply using msg, wrong outstanding message check, missing
    user failure handler configuration).


Changes since 1.1.5
~~~~~~~~~~~~~~~~~~~

* [fix] Updating documentation.

* [fix] Fixed vortex-regression-client.c to properly check and run on
  windows. Several updates on windows files to produce the bundle.

* [fix] Making vortex_gethostbyname to not break when receiving NULL
  (which also avoids vortex_connection_sock_connect to not break when
  received NULL ctx).

* [fix] Fixed double deallocation error for ctx->profiles_list_mutex.  
  Reported and fixed by Benoit Amiaux. 

* [fix] Fixed warning at __vortex_connection_new which was returning
  axl_false instead of NULL. Reported and fixed by Benoit Amiaux.

* [fix] Fixed vortex_tls_get_digest_sized on win32 which was causing a
  wrong memory access due to wrong buffer size indication. Reported
  and fixed by Benoit Amiaux.

* [fix] Fixed double dealloction on receiving frame when the channel
  frame received or profile frame received invocation failed. Bug
  reported and provided patch by Benoit Amiaux. Added regression test
  test02-m3 to reproduce the bug and check the fix introduced.

* [fix] Making vortex connection creation to acquire and release a
  reference to the context to allow closing a connection (and release
  its resources) even after having called vortex_exit_ctx.  Added
  regression test test-01q to reproduce the bug and check the fix
  introduced.

* [fix] Making vortex connection internal close to avoid contacting
  vortex sequencer to drop pending packages when a vortex context exit
  is in place.

* [fix] Making vortex_channel_new to acquire an internal reference to
  the channel created to avoid races between this function and the on
  created handler. Checked in test_16 alive reg test. Making also to
  acquire a reference to the connection during the creation process.

* [fix] Making vortex thread pool to acquire independent references to
  func and user data pointers to allow releasing internal mutexes
  during user code invocation. Making thread pool module to skip user
  code invocation in a vortex exit is in place.

* [fix] General updates to vortex alive module to decouple channel
  creation from the alive check itself to avoid races that may cause
  the alive check to monopolize (which was integrating channel
  creation) resources. Fixed several wrong access. Many thanks to
  DEXTER which provided a patch that fixed several bugs in the initial
  code (wrong reply using msg, wrong outstanding messages check,
  missing user failure handler configuration).

* [fix] Fixed unused Ctx references when ENABLE_LOG_SUPPORT
  is not defined.

* [fix] More updates to alive regression test (test_16). No bug found.

* [fix] Implemented reference counting support for objects
  representing events to be executed avoid races.

* [fix] Updated regression test to improve alive checkings.

* [fix] Adding initial vortex.alive PyVortex implementation.

* [new] Updated PyVortex connection module to include two new methods:
  blocked and is_blocked as bindings for its associated C API.

* [fix] More updates to PyVortex alive API. Added initial reg tests.

* [fix] Making vortex channel pool to acquire a reference to channels
  added allowing to check their internal status even after having the
  channel closed. Checked test_03-a. This also fixes test_22 inside
  pyvortex which was warning when run under valgrind.

* [fix] Updated internal channel pool same session check to test if
  ctx reference is defined to acquire it from the pool (avoids nasty
  context undefined message).

* [fix] Fixed channel pool race inside PyVortex when configured on
  created channel pool handler which was causing to nullify internal
  reference if the thread planner chooses to notify pool and then
  complete caller code (ouch!!)

* [fix] Adding more documentation to PyVortex.

* [fix] Added additional checks to alive tests (test_16) to ensure
  failure handlers is properly triggered.

* [fix] Improved alive internal implementation to also check
  connection activity to avoid triggering alive failure only based on
  its conditions. This better supports introducing alive on
  connections with heavy traffic (avoiding closing the connection
  because alive checks do not progress).

* [new] Updating debian packaging to produce new packages for alive
  implementation (including python binding).

* [fix] Adding PyVortex alive documentation.

* [new] Added initial implementation of vortex.Handle inside PyVortex
  API which will allow wrapping pointers used by the PyVortex
  API. Used this new object to implement selective connection on close
  removal. API added:

   - vortex.Connection.remove_on_close

* [fix] Committing missing alive.py file..

* [fix] Updates to libvortex-1.1.def and auxiliary files.

* [fix] Added regression test to check PyVortex
  vortex.Connection.remove_on_close method. Added documentation for
  this method.

* [fix] Fixed PyVortex vortex.AsyncQueue.timedout which was not
  working at all (fixed proper parameter parsing and added support to
  remove None in case of NULL received).

* [fix] Making PyVortexHandle destroy data function to unlock thread
  and reacquire it then user space call is finished.

* [fix] Fixed memory leak at vortex.alive implementation when a
  failure handler was configured.

* [fix] Making several vortex.Connection attributes at the PyVortex
  implementation to return None in case they are not defined.

* [fix] Making vortex_connection_new internal implementation to check
  if the VortexCtx provided is finishing to avoid creating a
  connection that will be in the middle. Added test_01-r to reproduce
  and check fix introduced.

* [fix] Making vortex channel close process to check if VortexCtx is
  being finished to unlock the wait for pending replies on the
  provided channel. Allows to call to finish vortex from inside frame
  received.  Added test-01-r to reproduce and check fix introduced.

* [fix] Making thread pool to not check for events in the case
  VortexCtx associated is being finished. Added test-01-r to reproduce
  and check fix introduced.

* [fix] Making vortex thread pool to acquire a reference to the
  VortexCtx it uses and to release it after exiting to avoid memory
  races.  Added test-01-r to reproduce and check fix introduced.

* [fix] Making vortex connection close notification to be protected by
  the internal mutex before calling user code. Updated current
  test_01r reg test to check this issue. All reg tests passed.

* [fix] Making vortex connection close notification to skip user
  handler calling in the case vortex context that holds the connection
  is being finished (due to a vortex_exit_ctx call).

* [fix] Added pthread_detach to vortex_thread_destroy in the case
  pthreads signals a deadlock after calling to pthread_join.

* [new] Updated vortex connection API to include a function that only
  shutdowns the connection socket (transport). Added new regression
  test to reproduce and introduce a fix to avoid locking the vortex
  reader in the case a connection close that never ends is installed.
  Still not able to reproduce the bug. Test 01-s.

* [fix] Making internal vortex connection close notification to use an
  async notification to the thread pool, avoiding to lock important
  threads like vortex reader and vortex sequencer when a connection
  close with a blocking task is called. Added regression test
  (test_01-s) to reproduce the bug and to check the fix introduced.

* [fix] Making vortex reader stop and vortex sequencer stop to have a
  timeout to avoid waiting for ever (60 seconds).

* [fix] Fixed minors unused variables when console debug is disabled.

* [fix] Minor timeout modification at test_01s

* [fix] Updated ALIVE and PyVortex ALIVE documentation.

* [fix] Fixed bug at ALIVE implementation which was not recording
  check period provided by the user, causing the check period to be 0.

* [fix] Fixed PyVortex Makefile.am which was not installing auxiliary
  files required to use vortex.alive.

About Us
~~~~~~~~

  Advanced Software Production Line provides GNU/Linux support and
  consulting services to help organizations to introduce GNU/Linux
  inside its process.

  Contact us, using English or Spanish, to get commercial support
  and/or BEEP based development services.

  You can reach us:

       http://www.aspl.es - info at aspl.es

  Enjoy Vortex Library!  

--
Francis Brosnan Blázquez           - francis at aspl.es
Advanced Software Production Line  - http://www.aspl.es
26th oct 2010, Madrid (Spain)




More information about the Vortex mailing list