=======================
HISTORY: MPI for Python
=======================
:Author: Lisandro Dalcin
:Contact: dalcinl@gmail.com
:Web Site: http://mpi4py.googlecode.com/
:Organization: CIMEC <http://www.cimec.org.ar>
:Address: CCT CONICET, 3000 Santa Fe, Argentina
Release 1.2.2 [2010-09-13]
==========================
* Add ``mpi4py.get_config()`` to retrieve information (compiler
wrappers, includes, libraries, etc) about the MPI implementation
employed to build mpi4py.
* Workaround Python libraries with missing GILState-related API calls
in case of non-threaded Python builds.
* Windows: look for MPICH2, DeinoMPI, Microsoft HPC Pack at their
default install locations under %ProgramFiles.
* MPE: fix hacks related to old API's, these hacks are broken when MPE
is built with a MPI implementations other than MPICH2.
* HP-MPI: fix for missing Fortran datatypes, use dlopen() to load the
MPI shared library before MPI_Init()
* Many distutils-related fixes, cleanup, and enhancements, better
logics to find MPI compiler wrappers.
* Support for ``pip install mpi4py``.
Release 1.2.1 [2010-02-26]
==========================
* Fix declaration in Cython include file. This declaration, while
valid for Cython, broke the simple-minded parsing used in
conf/mpidistutils.py to implement configure-tests for availability
of MPI symbols.
* Update SWIG support and make it compatible with Python 3. Also
generate an warning for SWIG < 1.3.28.
* Fix distutils-related issues in Mac OS X. Now ARCHFLAGS environment
variable is honored of all Python's ``config/Makefile`` variables.
* Fix issues with Open MPI < 1.4.2 releated to error checking and
``MPI_XXX_NULL`` handles.
Release 1.2 [2009-12-29]
========================
* Automatic MPI datatype discovery for NumPy arrays and PEP-3118
buffers. Now buffer-like objects can be messaged directly, it is no
longer required to explicitly pass a 2/3-list/tuple like ``[data,
MPI.DOUBLE]``, or ``[data, count, MPI.DOUBLE]``. Only basic types
are supported, i.e., all C/C99-native signed/unsigned integral types
and single/double precision real/complex floating types. Many thanks
to Eilif Muller for the initial feedback.
* Nonblocking send of pickled Python objects. Many thanks to Andreas
Kloeckner for the initial patch and enlightening discussion about
this enhancement.
* ``Request`` instances now hold a reference to the Python object
exposing the buffer involved in point-to-point communication or
parallel I/O. Many thanks to Andreas Kloeckner for the initial
feedback.
* Support for logging of user-defined states and events using `MPE
<http://www.mcs.anl.gov/research/projects/perfvis/>`_. Runtime
(i.e., without requiring a recompile!) activation of logging of all
MPI calls is supported in POSIX platforms implementing ``dlopen()``.
* Support for all the new features in MPI-2.2 (new C99 and F90
datatypes, distributed graph topology, local reduction operation,
and other minor enhancements).
* Fix the annoying issues related to Open MPI and Python dynamic
loading of extension modules in platforms supporting ``dlopen()``.
* Fix SLURM dynamic loading issues on SiCortex. Many thanks to Ian
Langmore for providing me shell access.
Release 1.1.0 [2009-06-06]
==========================
* Fix bug in ``Comm.Iprobe()`` that caused segfaults as Python C-API
calls were issued with the GIL released (issue #2).
* Add ``Comm.bsend()`` and ``Comm.ssend()`` for buffered and
synchronous send semantics when communicating general Python
objects.
* Now the call ``Info.Get(key)`` return a *single* value (i.e, instead
of a 2-tuple); this value is ``None`` if ``key`` is not in the
``Info`` object, or a string otherwise. Previously, the call
redundantly returned ``(None, False)`` for missing key-value pairs;
``None`` is enough to signal a missing entry.
* Add support for parametrized Fortran datatypes.
* Add support for decoding user-defined datatypes.
* Add support for user-defined reduction operations on memory
buffers. However, at most 16 user-defined reduction operations
can be created. Ask the author for more room if you need it.
Release 1.0.0 [2009-03-20]
==========================
This is the fist release of the all-new, Cython-based, implementation
of *MPI for Python*. Unfortunately, this implementation is not
backward-compatible with the previous one. The list below summarizes
the more important changes that can impact user codes.
* Some communication calls had *overloaded* functionality. Now there
is a clear distinction between communication of general Python
object with *pickle*, and (fast, near C-speed) communication of
buffer-like objects (e.g., NumPy arrays).
- for communicating general Python objects, you have to use
all-lowercase methods, like ``send()``, ``recv()``, ``bcast()``,
etc.
- for communicating array data, you have to use ``Send()``,
``Recv()``, ``Bcast()``, etc. methods. Buffer arguments to these
calls must be explicitly specified by using a 2/3-list/tuple like
``[data, MPI.DOUBLE]``, or ``[data, count, MPI.DOUBLE]`` (the
former one uses the byte-size of ``data`` and the extent of the
MPI datatype to define the ``count``).
* Indexing a communicator with an integer returned a special object
associating the communication with a target rank, alleviating you
from specifying source/destination/root arguments in point-to-point
and collective communications. This functionality is no longer
available, expressions like::
MPI.COMM_WORLD[0].Send(...)
MPI.COMM_WORLD[0].Recv(...)
MPI.COMM_WORLD[0].Bcast(...)
have to be replaced by::
MPI.COMM_WORLD.Send(..., dest=0)
MPI.COMM_WORLD.Recv(..., source=0)
MPI.COMM_WORLD.Bcast(..., root=0)
* Automatic MPI initialization (i.e., at import time) requests the
maximum level of MPI thread support (i.e., it is done by calling
``MPI_Init_thread()`` and passing ``MPI_THREAD_MULTIPLE``). In case
you need to change this behavior, you can tweak the contents of the
``mpi4py.rc`` module.
* In order to obtain the values of predefined attributes attached to
the world communicator, now you have to use the ``Get_attr()``
method on the ``MPI.COMM_WORLD`` instance::
tag_ub = MPI.COMM_WORLD.Get_attr(MPI.TAG_UB)
* In the previous implementation, ``MPI.COMM_WORLD`` and
``MPI.COMM_SELF`` were associated to **duplicates** of the (C-level)
``MPI_COMM_WORLD`` and ``MPI_COMM_SELF`` predefined communicator
handles. Now this is no longer the case, ``MPI.COMM_WORLD`` and
``MPI.COMM_SELF`` proxies the **actual** ``MPI_COMM_WORLD`` and
``MPI_COMM_SELF`` handles.
* Convenience aliases ``MPI.WORLD`` and ``MPI.SELF`` were removed. Use
instead ``MPI.COMM_WORLD`` and ``MPI.COMM_SELF``.
* Convenience constants ``MPI.WORLD_SIZE`` and ``MPI.WORLD_RANK`` were
removed. Use instead ``MPI.COMM_WORLD.Get_size()`` and
``MPI.COMM_WORLD.Get_rank()``.
