1 Introduction
2 ------------
3
4 The pprocess module provides elementary support for parallel programming in
5 Python using a fork-based process creation model in conjunction with a
6 channel-based communications model implemented using socketpair and poll. On
7 systems with multiple CPUs or multicore CPUs, processes should take advantage
8 of as many CPUs or cores as the operating system permits.
9
10 Tutorial
11 --------
12
13 The tutorial provides some information about the examples described below.
14 See the docs/tutorial.html file in the distribution for more details.
15
16 Reference
17 ---------
18
19 A description of the different mechanisms provided by the pprocess module can
20 be found in the reference document. See the docs/reference.html file in the
21 distribution for more details.
22
23 Quick Start
24 -----------
25
26 Try running the simple examples. For example:
27
28 PYTHONPATH=. python examples/simple_create.py
29
30 (These examples show in different ways how limited number of processes can be
31 used to perform a parallel computation. The simple.py, simple1.py, simple2.py
32 and simple_map.py programs are sequential versions of the other programs.)
33
34 The following table summarises the features used in the programs:
35
36 Program (.py) pmap MakeParallel manage start create Map Queue Exchange
37 ------------- ---- ------------ ------ ----- ------ --- ----- --------
38 simple_create_map Yes Yes
39 simple_create_queue Yes Yes
40 simple_create Yes Yes
41 simple_managed_map Yes Yes Yes
42 simple_managed_queue Yes Yes Yes
43 simple_managed Yes Yes Yes
44 simple_pmap Yes
45 simple_pmap_iter Yes
46 simple_start_queue Yes Yes Yes
47 simple_start Yes Yes
48
49 The simplest parallel programs are simple_pmap.py and simple_pmap_iter.py
50 which employ the pmap function resembling the built-in map function in
51 Python.
52
53 Other simple programs are those employing the Queue class, together with those
54 using the manage method which associates functions or callables with Queue or
55 Exchange objects for convenient invocation of those functions and the
56 management of their communications.
57
58 The most technically involved program is simple_start.py which uses the
59 Exchange class together with a calculation function which is aware of the
60 parallel environment and which communicates over the supplied communications
61 channel directly to the creating process.
62
63 It should be noted that with the exception of simple_start.py, those examples
64 employing calculation functions (as opposed to doing a calculation inline in a
65 loop body) all use MakeParallel to make those functions parallel-aware, thus
66 permitting the conversion of "normal" functions to a form usable in the
67 parallel environment.
68
69 Reusable Processes
70 ------------------
71
72 An additional example not listed above, simple_managed_map_reusable.py,
73 employs the MakeReusable class instead of MakeParallel in order to demonstrate
74 reusable processes and channels:
75
76 PYTHONPATH=. python examples/simple_managed_map_reusable.py
77
78 Persistent Processes
79 --------------------
80
81 A number of persistent variants of some of the above examples employ a
82 persistent or background process which can be started by one process and
83 contacted later by another in order to collect the results of a computation.
84 For example:
85
86 PYTHONPATH=. python examples/simple_persistent_managed.py --start
87 PYTHONPATH=. python examples/simple_persistent_managed.py --reconnect
88
89 PYTHONPATH=. python examples/simple_background_queue.py --start
90 PYTHONPATH=. python examples/simple_background_queue.py --reconnect
91
92 PYTHONPATH=. python examples/simple_persistent_queue.py --start
93 PYTHONPATH=. python examples/simple_persistent_queue.py --reconnect
94
95 Parallel Raytracing with PyGmy
96 ------------------------------
97
98 The PyGmy raytracer modified to use pprocess can be run to investigate the
99 potential for speed increases in "real world" programs:
100
101 cd examples/PyGmy
102 PYTHONPATH=../..:. python scene.py
103
104 (This should produce a file called test.tif - a TIFF file containing a
105 raytraced scene image.)
106
107 Test Programs
108 -------------
109
110 There are some elementary tests:
111
112 PYTHONPATH=. python tests/create_loop.py
113 PYTHONPATH=. python tests/start_loop.py
114
115 (Simple loop demonstrations which use two different ways of creating and
116 starting the parallel processes.)
117
118 PYTHONPATH=. python tests/start_indexer.py <directory>
119
120 (A text indexing demonstration, where <directory> should be a directory
121 containing text files to be indexed, although HTML files will also work well
122 enough. After indexing the files, a prompt will appear, words or word
123 fragments can be entered, and matching words and their locations will be
124 shown. Run the program without arguments to see more information.)
125
126 Contact, Copyright and Licence Information
127 ------------------------------------------
128
129 The current Web page for pprocess at the time of release is:
130
131 http://www.boddie.org.uk/python/pprocess.html
132
133 The author can be contacted at the following e-mail address:
134
135 paul@boddie.org.uk
136
137 Copyright and licence information can be found in the docs directory - see
138 docs/COPYING.txt, docs/lgpl-3.0.txt and docs/gpl-3.0.txt for more information.
139
140 For the PyGmy raytracer example, different copyright and licence information
141 is provided in the docs directory - see docs/COPYING-PyGmy.txt and
142 docs/LICENCE-PyGmy.txt for more information.
143
144 Dependencies
145 ------------
146
147 This software depends on standard library features which are stated as being
148 available only on "UNIX"; it has only been tested repeatedly on a GNU/Linux
149 system, and occasionally on systems running OpenSolaris.
150
151 New in pprocess 0.4.1 (Changes since pprocess 0.4)
152 --------------------------------------------------
153
154 * Fixed the get_number_of_cores function to work with /proc/cpuinfo where
155 the "physical id" field is missing.
156 * Changed the Map class to permit incremental access to received results
157 from completed parts of the sequence of inputs, also adding an iteration
158 interface.
159 * Added an example, simple_pmap_iter.py, to demonstrate iteration over maps.
160
161 New in pprocess 0.4 (Changes since pprocess 0.3.1)
162 --------------------------------------------------
163
164 * Added support for persistent/background processes.
165 * Added a utility function to detect and return the number of processor
166 cores available.
167 * Added missing documentation stylesheet.
168 * Added support for Solaris using pipes instead of socket pairs, since
169 the latter do not apparently work properly with poll on Solaris.
170
171 New in pprocess 0.3.1 (Changes since pprocess 0.3)
172 --------------------------------------------------
173
174 * Moved the reference material out of the module docstring and into a
175 separate document, converting it to XHTML in the process.
176 * Fixed the project name in the setup script.
177
178 New in pprocess 0.3 (Changes since parallel 0.2.5)
179 --------------------------------------------------
180
181 * Added managed callables: wrappers around callables which cause them to be
182 automatically managed by the exchange from which they were acquired.
183 * Added MakeParallel: a wrapper instantiated around a normal function which
184 sends the result of that function over the supplied channel when invoked.
185 * Added MakeReusable: a wrapper like MakeParallel which can be used in
186 conjunction with the newly-added reuse capability of the Exchange class in
187 order to reuse processes and channels.
188 * Added a Map class which attempts to emulate the built-in map function,
189 along with a pmap function using this class.
190 * Added a Queue class which provides a simpler iterator-style interface to
191 data produced by created processes.
192 * Added a create method to the Exchange class and an exit convenience
193 function to the module.
194 * Changed the Exchange implementation to not block when attempting to start
195 new processes beyond the process limit: such requests are queued and
196 performed as running processes are completed. This permits programs using
197 the start method to proceed to consumption of results more quickly.
198 * Extended and updated the examples. Added a tutorial.
199 * Added Ubuntu Feisty (7.04) package support.
200
201 New in parallel 0.2.5 (Changes since parallel 0.2.4)
202 ----------------------------------------------------
203
204 * Added a start method to the Exchange class for more convenient creation of
205 processes.
206 * Relicensed under the LGPL (version 3 or later) - this also fixes the
207 contradictory situation where the GPL was stated in the pprocess module
208 (which was not, in fact, the intention) and the LGPL was stated in the
209 documentation.
210
211 New in parallel 0.2.4 (Changes since parallel 0.2.3)
212 ----------------------------------------------------
213
214 * Set buffer sizes to zero for the file object wrappers around sockets: this
215 may prevent deadlock issues.
216
217 New in parallel 0.2.3 (Changes since parallel 0.2.2)
218 ----------------------------------------------------
219
220 * Added convenient message exchanges, offering methods handling common
221 situations at the cost of having to define a subclass of Exchange.
222 * Added a simple example of performing a parallel computation.
223 * Improved the PyGmy raytracer example to use the newly added functionality.
224
225 New in parallel 0.2.2 (Changes since parallel 0.2.1)
226 ----------------------------------------------------
227
228 * Changed the status testing in the Exchange class, potentially fixing the
229 premature closure of channels before all data was read.
230 * Fixed the PyGmy raytracer example's process accounting by relying on the
231 possibly more reliable Exchange behaviour, whilst also preventing
232 erroneous creation of "out of bounds" processes.
233 * Added a removed attribute on the Exchange to record which channels were
234 removed in the last call to the ready method.
235
236 New in parallel 0.2.1 (Changes since parallel 0.2)
237 --------------------------------------------------
238
239 * Added a PyGmy raytracer example.
240 * Updated copyright and licensing details (FSF address, additional works).
241
242 New in parallel 0.2 (Changes since parallel 0.1)
243 ------------------------------------------------
244
245 * Changed the name of the included module from parallel to pprocess in order
246 to avoid naming conflicts with PyParallel.
247
248 Release Procedures
249 ------------------
250
251 Update the pprocess __version__ attribute and the setup.py file version field.
252 Change the version number and package filename/directory in the documentation.
253 Update the release notes (see above).
254 Check the release information in the PKG-INFO file.
255 Tag, export.
256 Archive, upload.
257 Update PyPI.
258
259 Making Packages
260 ---------------
261
262 To make Debian-based packages:
263
264 1. Create new package directories under packages if necessary.
265 2. Make a symbolic link in the distribution's root directory to keep the
266 Debian tools happy:
267
268 ln -s packages/ubuntu-hoary/python2.4-parallel-pprocess/debian/
269
270 Or:
271
272 ln -s packages/ubuntu-feisty/python-pprocess/debian/
273
274 3. Run the package builder:
275
276 dpkg-buildpackage -rfakeroot
277
278 4. Locate and tidy up the packages in the parent directory of the
279 distribution's root directory.