blob: f183853f35a8596b072f82ca3bf73d71e3bbb3e5 [file] [log] [blame]
Andrew MacIntyre41d97d62002-02-17 05:23:30 +00001This is a port of Python 2.3 to OS/2 using the EMX development tools
2=========================================================================
3
4What's new since the previous release
5-------------------------------------
6
7This release of the port incorporates the following changes from the
Andrew MacIntyre7e5bf672002-12-04 12:40:49 +00008October 24, 2002 release of the Python 2.2.2 port:
Andrew MacIntyre41d97d62002-02-17 05:23:30 +00009
10- based on the Python v2.3 final release source.
Andrew MacIntyreffcf8992002-08-18 06:28:21 +000011- now setting higher number of file handles (250).
12- defaults to building with PyMalloc enabled (Python 2.3 default).
13- the port is now maintained in the Python CVS repository.
14
15Python 2.3 incorporates several changes which have resolved the
16longstanding problems the EMX port has had with test_longexp (used
17to be "YOU HAVE BEEN WARNED" item 1).
Andrew MacIntyre41d97d62002-02-17 05:23:30 +000018
19
20Licenses and info about Python and EMX
21--------------------------------------
22
23Please read the file README.Python-2.3 included in this package for
24information about Python 2.3. This file is the README file from the
25Python 2.3 source distribution available via http://www.python.org/
26and its mirrors. The file LICENCE.Python-2.3 is the text of the Licence
27from the Python 2.3 source distribution.
28
29Note that the EMX package that this package depends on is released under
30the GNU General Public Licence. Please refer to the documentation
31accompanying the EMX Runtime libraries for more information about the
32implications of this. A copy of version 2 of the GPL is included as the
33file COPYING.gpl2.
34
35Readline and GDBM are covered by the GNU General Public Licence. I think
36Eberhard Mattes' porting changes to BSD DB v1.85 are also GPL'ed (BSD DB
37itself is BSD Licenced). ncurses and expat appear to be covered by MIT
38style licences - please refer to the source distributions for more detail.
39zlib is distributable under a very free license. GNU MP and GNU UFC are
40under the GNU LGPL (see file COPYING.lib).
41
42My patches to the Python-2.x source distributions, and any other packages
43used in this port, are placed in the public domain.
44
45This software is provided 'as-is', without any express or implied warranty.
46In no event will the author be held liable for any damages arising from the
47use of the software.
48
49I do hope however that it proves useful to someone.
50
51
52Other ports
53-----------
54
55There have been ports of previous versions of Python to OS/2.
56
57The best known would be that by Jeff Rush, most recently of version
581.5.2. Jeff used IBM's Visual Age C++ (v3) for his ports, and his
59patches have been included in the Python 2.3 source distribution.
60
61Andrew Zabolotny implemented a port of Python v1.5.2 using the EMX
62development tools. His patches against the Python v1.5.2 source
63distribution have become the core of this port, and without his efforts
64this port wouldn't exist. Andrew's port also appears to have been
65compiled with his port of gcc 2.95.2 to EMX, which I have but have
66chosen not to use for the binary distribution of this port (see item 21
67of the "YOU HAVE BEEN WARNED" section below).
68
69Previous Python port releases by me:-
70 - v2.0 on March 31, 2001;
71 - v2.0 on April 25, 2001 (cleanup release + Stackless variant);
72 - v2.1 on June 17, 2001;
73 - v2.0 (Stackless re-release) on June 18, 2001.
74 - v2.1.1 on August 5, 2001;
75 - v2.1.1 on August 12, 2001 (cleanup release);
Andrew MacIntyreffcf8992002-08-18 06:28:21 +000076 - v2.1.1 (updated DLL) on August 14, 2001;
77 - v2.2b2 on December 8, 2001 (not uploaded to archive sites);
78 - v2.2c1 on December 16, 2001 (not uploaded to archive sites);
79 - v2.2 on December 24, 2001;
80 - v2.2.1c2 on March 31, 2002 (not uploaded to archive sites);
81 - v2.2.1 on April 14, 2002.
Andrew MacIntyre7e5bf672002-12-04 12:40:49 +000082 - v2.2.2 on October 24, 2002.
Andrew MacIntyre41d97d62002-02-17 05:23:30 +000083
84It is possible to have these earlier ports still usable after installing
85this port - see the README.os2emx.multiple_versions file, contributed by
86Dr David Mertz, for a suggested approach to achieving this.
87
88
89Software requirements
90---------------------
91
92This package requires the EMX Runtime package, available from the
93Hobbes (http://hobbes.nmsu.edu/) and LEO (http://archiv.leo.org/)
94archives of OS/2 software. I have used EMX version 0.9d fix04 in
95developing this port.
96
97My development system is running OS/2 v4 with fixpack 12.
98
993rd party software which has been linked into dynamically loaded modules:
100- ncurses (see http://dickey.his.com/ for more info, v5.2)
101- GNU Readline (Kai Uwe Rommel's port available from Hobbes or LEO, v2.1)
102- GNU GDBM (Kai Uwe Rommel's port available from Hobbes or LEO, v1.7.3)
103- zlib (Hung-Chi Chu's port available from Hobbes or LEO, v1.1.3)
Andrew MacIntyreffcf8992002-08-18 06:28:21 +0000104- expat (distributed with Python, v1.95.2)
Andrew MacIntyre41d97d62002-02-17 05:23:30 +0000105- GNU MP (Peter Meerwald's port available from LEO, v2.0.2)
106- GNU UFC (Kai Uwe Rommel's port available from LEO, v2.0.4)
107
108The zlib module requires the Z.DLL to be installed - see the Installation
109section and item 12 of the "YOU HAVE BEEN WARNED" section for more
110information.
111
112About this port
113---------------
114
115I have attempted to make this port as complete and functional as I can,
116notwithstanding the issues in the "YOU HAVE BEEN WARNED" section below.
117
118Core components:
119
120Python.exe is linked as an a.out executable, ie using EMX method E1
121to compile & link the executable. This is so that fork() works (see
122"YOU HAVE BEEN WARNED" item 2).
123
124Python23.dll is created as a normal OMF DLL, with an OMF import
125library and module definition file. There is also an a.out (.a) import
Andrew MacIntyreffcf8992002-08-18 06:28:21 +0000126library to support linking the DLL to a.out executables. The DLL
127requires the EMX runtime DLLs.
Andrew MacIntyre41d97d62002-02-17 05:23:30 +0000128
129This port has been built with complete support for multithreading.
130
131Modules:
132
133As far as possible, extension modules have been made dynamically loadable
134when the module is intended to be built this way. I haven't yet changed
135the building of Python's standard modules over to using the DistUtils.
136
137See "YOU HAVE BEEN WARNED" item 5 for notes about the fcntl module, and
138"YOU HAVE BEEN WARNED" item 14 for notes about the pwd and grp modules.
139
140Support for case sensitive module import semantics has been added to match
141the Windows release. This can be deactivated by setting the PYTHONCASEOK
142environment variable (the value doesn't matter) - see "YOU HAVE BEEN WARNED"
143item 16.
144
145Optional modules:
146
147Where I've been able to locate the required 3rd party packages already
148ported to OS/2, I've built and included them.
149
150These include ncurses (_curses, _curses_panel), BSD DB (bsddb),
151GNU GDBM (gdbm, dbm), zlib (zlib), GNU Readline (readline), expat
152(pyexpat), GNU MP (mpz) and GNU UFC (crypt).
153
154I have built these modules statically linked against the 3rd party
155libraries, with the exception of zlib. Unfortunately my attempts to use
156the dll version of GNU readline have been a dismal failure, in that when
157the dynamically linked readline module is active other modules
158immediately provoke a core dump when imported.
159
160Only the BSD DB package (part of the BSD package distributed with EMX)
161needed source modifications to be used for this port, pertaining to use
162of errno with multithreading.
163
164The other packages, except for ncurses and zlib, needed Makefile changes
165for multithreading support but no source changes.
166
167The _curses_panel module is a potential problem - see "YOU HAVE BEEN
168WARNED" item 17.
169
170Upstream source patches:
171
172No updates to the Python 2.3 release have become available.
173
174Eberhard Mattes' EMXFIX04 update to his EMX 0.9d tools suite includes
175bug fixes for the BSD DB library. The bsddb module included in this
176port incorporates these fixes.
177
178Library and other distributed Python code:
179
180The Python standard library lives in the Lib directory. All the standard
181library code included with the Python 2.3 source distribution is included
182in the binary archive, with the exception of the dos-8x3 and tkinter
183subdirectories which have been omitted to reduce the size of the binary
184archive - the dos-8x3 components are unnecessary duplicates and Tkinter
185is not supported by this port (yet). All the plat-* subdirectories in the
186source distribution have also been omitted, and a plat-os2emx directory
187included.
188
189The Tools and Demo directories contain a collection of Python scripts.
190To reduce the size of the binary archive, the Demo/sgi, Demo/Tix,
191Demo/tkinter, Tools/audiopy and Tools/IDLE subdirectories have been
192omitted as not being supported by this port. The Misc directory has
193also been omitted.
194
195All subdirectories omitted from the binary archive can be reconstituted
196from the Python 2.3 source distribution, if desired.
197
198Support for building Python extensions:
199
200The Config subdirectory contains the files describing the configuration
201of the interpreter and the Makefile, import libraries for the Python DLL,
202and the module definition file used to create the Python DLL. The
203Include subdirectory contains all the standard Python header files
204needed for building extensions.
205
206As I don't have the Visual Age C++ compiler, I've made no attempt to
207have this port support extensions built with that compiler.
208
209
210Packaging
211---------
212
213This port is packaged into several archives:
214- python-2.3-os2emx-bin-02????.zip (binaries, library modules)
215- python-2.3-os2emx-src-03????.zip (source patches and makefiles)
216
217Documentation for the Python language, as well as the Python 2.3
218source distibution, can be obtained from the Python website
219(http://www.python.org/) or the Python project pages at Sourceforge
220(http://sf.net/projects/python/).
221
222
223Installation
224------------
225
226Obtain and install, as per the included instructions, the EMX runtime
227package.
228
229If you wish to use the zlib module, you will need to obtain and install
230the Z.DLL from Hung-Chi Chu's port of zlib v1.1.3 (zlib113.zip). See also
231"YOU HAVE BEEN WARNED" item 12 below.
232
233Unpack this archive, preserving the subdirectories, in the root directory
234of the drive where you want Python to live.
235
236Add the Python directory (eg C:\Python23) to the PATH and LIBPATH
237variables in CONFIG.SYS.
238
239You should then set the PYTHONHOME and PYTHONPATH environment variables
240in CONFIG.SYS.
241
242PYTHONHOME should be set to Python's top level directory. PYTHONPATH
243should be set to the semicolon separated list of principal Python library
244directories.
245I use:
246 SET PYTHONHOME=F:/Python23
247 SET PYTHONPATH=F:/Python23/Lib;F:/Python23/Lib/plat-os2emx;
248 F:/Python23/Lib/lib-dynload;F:/Python23/Lib/site-packages
249
250NOTE!: the PYTHONPATH setting above is linewrapped for this document - it
251should all be on one line in CONFIG.SYS!
252
253If you wish to use the curses module, you should set the TERM and TERMINFO
254environment variables appropriately.
255
256If you don't already have ncurses installed, I have included a copy of the
257EMX subset of the Terminfo database included with the ncurses-5.2 source
258distribution. This can be used by setting the TERMINFO environment variable
259to the path of the Terminfo subdirectory below the Python home directory.
260On my system this looks like:
261 SET TERMINFO=F:/Python23/Terminfo
262
263For the TERM environment variable, I would try one of the following:
264 SET TERM=ansi
265 SET TERM=os2
266 SET TERM=window
267
268You will have to reboot your system for these changes to CONFIG.SYS to take
269effect.
270
271If you wish to compile all the included Python library modules to bytecode,
272you can change into the Python home directory and run the COMPILEALL.CMD
273batch file.
274
275You can execute the regression tests included with the Python 2.3 source
276distribution by changing to the Python 2.3 home directory and executing the
277REGRTEST.CMD batch file. The following tests are known to fail at this
278time:
279- test_longexp (see "YOU HAVE BEEN WARNED" item 1);
280- test_mhlib (I don't know of any port of MH to OS/2);
281- test_pwd (see "YOU HAVE BEEN WARNED" item 14, probably a bug in my code);
282- test_grp (as per test_pwd);
283- test_strftime (see "YOU HAVE BEEN WARNED" item 20);
284- test_socketserver (fork() related, see "YOU HAVE BEEN WARNED" item 2).
285
286
287YOU HAVE BEEN WARNED!!
288----------------------
289
290I know about a number of nasties in this port.
291
Andrew MacIntyreffcf8992002-08-18 06:28:21 +0000292{1. Issue resolved...}
Andrew MacIntyre41d97d62002-02-17 05:23:30 +0000293
2942. Eberhard Mattes, author of EMX, writes in his documentation that fork()
295is very inefficient in the OS/2 environment. It also requires that the
296executable be linked in a.out format rather than OMF. Use the os.exec
297and/or the os.spawn family of functions where possible.
298
299{3. Issue resolved...}
300
3014. In the absence of GNU Readline, terminating the interpreter requires a
302control-Z (^Z) followed by a carriage return. Jeff Rush documented this
303problem in his Python 1.5.2 port. With Readline, a control-D (^D) works
304as per the standard Unix environment.
305
3065. EMX only has a partial implementation of fcntl(). The fcntl module
307in this port supports what EMX supports. If fcntl is important to you,
308please review the EMX C Library Reference (included in .INF format in the
309EMXVIEW.ZIP archive as part of the complete EMX development tools suite).
310Because of other side-effects I have modified the test_fcntl.py test
311script to deactivate the exercising of the missing functionality.
312
3136. The BSD DB module is linked against DB v1.85. This version is widely
314known to have bugs, although some patches have become available (and are
315incorporated into the included bsddb module). Unless you have problems
316with software licenses which would rule out GDBM (and the dbm module
317because it is linked against the GDBM library) or need it for file format
318compatibility, you may be better off deleting it and relying on GDBM. I
319haven't looked at porting the version of the module supporting the later
320SleepyCat releases of BSD DB, which would also require a port of the
321SleepyCat DB package.
322
3237. The readline module has been linked against ncurses rather than the
324termcap library supplied with EMX.
325
326{8. Workaround implemented}
327
3289. I have configured this port to use "/" as the preferred path separator
329character, rather than "\" ('\\'), in line with the convention supported
330by EMX. Backslashes are still supported of course, and still appear in
331unexpected places due to outside sources that don't get normalised.
332
33310. While the DistUtils components are now functional, other
334packaging/binary handling tools and utilities such as those included in
335the Demo and Tools directories - freeze in particular - are unlikely to
336work. If you do get them going, I'd like to know about your success.
337
33811. I haven't set out to support the [BEGIN|END]LIBPATH functionality
339supported by one of the earlier ports (Rush's??). If it works let me know.
340
34112. There appear to be several versions of Z.DLL floating around - the one
342I have is 45061 bytes and dated January 22, 1999. I have a report that
343another version causes SYS3175s when the zlib module is imported.
344
34514. As a result of the limitations imposed by EMX's library routines, the
346standard extension module pwd only synthesises a simple passwd database,
347and the grp module cannot be supported at all.
348
349I have written substitutes, in Python naturally, which can process real
350passwd and group files for those applications (such as MailMan) that
351require more than EMX emulates. I have placed pwd.py and grp.py in
352Lib/plat-os2emx, which is usually before Lib/lib-dynload (which contains
353pwd.pyd) in the PYTHONPATH. If you have become attached to what pwd.pyd
354supports, you can put Lib/lib-dynload before Lib/plat-os2emx in PYTHONPATH
355or delete/rename pwd.py & grp.py.
356
357pwd.py & grp.py support locating their data files by looking in the
358environment for them in the following sequence:
359pwd.py: $ETC_PASSWD (%ETC_PASSWD%)
360 $ETC/passwd (%ETC%/passwd)
361 $PYTHONHOME/Etc/passwd (%PYTHONHOME%/Etc/passwd)
362grp.py: $ETC_GROUP (%ETC_GROUP%)
363 $ETC/group (%ETC%/group)
364 $PYTHONHOME/Etc/group (%PYTHONHOME%/Etc/group)
365
366Both modules support using either the ":" character (Unix standard) or
367";" (OS/2, DOS, Windows standard) field separator character, and pwd.py
368implements the following drive letter conversions for the home_directory and
369shell fields (for the ":" separator only):
370 $x -> x:
371 x; -> x:
372
373Example versions of passwd and group are in the Etc subdirectory. Note
374that as of this release, this code fails the regression test. I'm looking
375into why, and hope to have this fixed.
376
37715. As of Python 2.1, termios support has mutated. There is no longer a
378platform specific TERMIOS.py containing the symbolic constants - these
379now live in the termios module. EMX's termios routines don't support all
380of the functionality now exposed by the termios module - refer to the EMX
381documentation to find out what is supported.
382
38316. The case sensitive import semantics introduced in Python 2.1 for other
384case insensitive but case preserving file/operating systems (Windows etc),
385have been incorporated into this port, and are active by default. Setting
386the PYTHONCASEOK environment variable (to any value) reverts to the
387previous (case insensitive) semantics.
388
38917. Because I am statically linking ncurses, the _curses_panel
390module has potential problems arising from separate library data areas.
391To avoid this, I have configured the _curses_.pyd (imported as
392"_curses_panel") to import the ncurses symbols it needs from _curses.pyd.
393As a result the _curses module must be imported before the _curses_panel
394module. As far as I can tell, the modules in the curses package do this.
395If you have problems attempting to use the _curses_panel support please
396let me know, and I'll look into an alternative solution.
397
39818. I tried enabling the Python Object Allocator (PYMALLOC) code. While
399the port built this way passes the regression test, the Numpy extension
400(I tested v19.0.0) as built with with the port's DistUtils code doesn't
401work. Specifically, attempting to "import Numeric" provokes a core dump.
402Supposedly Numpy v20.1.0 contains a fix for this, but for reason outlined
403in item 1 above, PYMALLOC is not enabled in this release.
404
40519. sys.platform now reports "os2emx" instead of "os2". os.name still
406reports "os2". This change was to make it easier to distinguish between
407the VAC++ build (being maintained by Michael Muller) and the EMX build
408(this port), principally for DistUtils.
409
41020. it appears that the %W substitution in the EMX strftime() routine has
411an off-by-one bug. strftime was listed as passing the regression tests
412in previous releases, but this fact appears to have been an oversight in
413the regression test suite. To fix this really requires a portable
414strftime routine - I'm looking into using one from FreeBSD, but its not
415ready yet.
416
41721. previous releases of my Python ports have used the GCC optimisations
418"-O2 -fomit-frame-pointer". After experimenting with various optimisation
419settings, including deactivating assert()ions, I have concluded that "-O2"
420appears the best compromise for GCC 2.8.1 on my hardware. Curiously,
421deactivating assert() (via defining NDEBUG) _negatively_ impacts
422performance, allbeit only slightly, so I've chosen to leave the assert()s
423active.
424
425I did try using Andrew Zabolotny's (p)gcc 2.95.2 compiler, and in
426general concluded that it produced larger objects that ran slower
427than Mattes' gcc 2.8.1 compiler.
428
429Pystone ratings varied from just over 2000/s (no optimisation at all)
430to just under 3300/s (gcc 2.8.1, -O2) on my K6/2-300 system, for
431100,000 iterations per run (rather than the default 10000).
432
433As a result of the optimisation change, the Python DLL is about 10%
434smaller than in the 2.1 release, and many of the dynamically loadable
435modules are smaller too.
436
437[2001/08/12]
438
43922. As of this release, os.spawnv() and os.spawnve() now expose EMX's
440library routines rather than use the emulation in os.py.
441
442In order to make use of some of the features this makes available in
443the OS/2 environment, you should peruse the relevant EMX documentation
444(EMXLIB.INF in the EMXVIEW.ZIP archive accompanying the EMX archives
445on Hobbes or LEO). Be aware that I have exposed all the "mode" options
446supported by EMX, but there are combinations that either cannot be
447practically used by/in Python or have the potential to compromise your
448system's stability.
449
45023. pythonpm.exe in previous releases was just python.exe with the
451WINDOWAPI linker option set in the pythonpm.def file. In practice,
452this turns out to do nothing useful.
453
454I have written a replacement which wraps the Python DLL in a genuine
455Presentation Manager application. This version actually runs the
456Python interpreter in a separate thread from the PM shell, in order
457that PythonPM has a functioning message queue as good PM apps should.
458In its current state, PythonPM's window is hidden. It can be displayed,
459although it will have no content as nothing is ever written to the
460window. Only the "hide" button is available. Although the code
461has support for shutting PythonPM down when the Python interpreter is
462still busy (via the "control" menu), this is not well tested and given
463comments I've come across in EMX documentation suggesting that the
464thread killing operation has problems I would suggest caution in
465relying on this capability.
466
467PythonPM processes commandline parameters normally. The standard input,
468output and error streams are only useful if redirected, as PythonPM's
469window is not a console in any form and so cannot accept or display
470anything. This means that the -i option is ineffective.
471
472Because the Python thread doesn't create its own message queue, creating
473PM Windows and performing most PM operations is not possible from within
474this thread. How this will affect supporting PM extensions (such as
475Tkinter using a PM port of Tcl/Tk, or wxPython using the PM port of
476WxWindows) is still being researched.
477
478Note that os.fork() _DOES_NOT_WORK_ in PythonPM - SYS3175s are the result
479of trying. os.spawnv() _does_ work. PythonPM passes all regression tests
480that the standard Python interpreter (python.exe) passes, with the exception
481of test_fork1 and test_socket which both attempt to use os.fork().
482
483I very much want feedback on the performance, behaviour and utility of
484PythonPM. I would like to add a PM console capability to it, but that
485will be a non-trivial effort. I may be able to leverage the code in
486Illya Vaes' Tcl/Tk port, which would make it easier.
487
488[2001/08/14]
489
49024. os.chdir() now uses EMX's _chdir2(), which supports changing
491both drive and directory at once. Similarly, os.getcwd() now uses
492EMX's _getcwd() which returns drive as well as path.
493
494[2001/12/08] - 2.2 Beta 2
495
49625. pyconfig.h (previously known as config.h) is now located in the
497Include subdirectory with all other include files.
498
499[2001/12/16] - 2.2 Release Candidate 1
500
501[2001/12/08] - 2.2 Final
502
Andrew MacIntyreffcf8992002-08-18 06:28:21 +0000503[2002/03/31] - 2.2.1 Release Candidate 2
504
505[2002/04/14] - 2.2.1 Final
506
507[2002/8/18]
508
50926. now explicitly set the number of file handles available to a
510Python process to 250. EMX default is 40, which is insufficient for the
511recently checked in security improvments to the tempfile regression
512test (test_tempfile) which tries to create 100 temporary files.
513
514This setting can be overridden via the EMXOPT environment variable:
515 set EMXOPT=-h250
516is equivalent to the setting currently used. The emxbind utility (if you
517have it installed) can also be used to permanently change the setting in
518python.exe - please refer to the EMX documentation for more information.
519
Andrew MacIntyre41d97d62002-02-17 05:23:30 +0000520... probably other issues that I've not encountered, or don't remember :-(
521
522If you encounter other difficulties with this port, which can be
523characterised as peculiar to this port rather than to the Python release,
524I would like to hear about them. However I cannot promise to be able to do
525anything to resolve such problems. See the Contact section below...
526
527
528To do...
529--------
530
531In no particular order of apparent importance or likelihood...
532
533- support Tkinter and/or alternative GUI (wxWindows??)
534
535
536Credits
537-------
538
539In addition to people identified above, I'd like to thank:
540- the BDFL, Guido van Rossum, and crew for Python;
541- Dr David Mertz, for trying out a pre-release of this port;
542- the Python-list/comp.lang.python community;
543- John Poltorak, for input about pwd/grp.
544
545Contact
546-------
547
548Constructive feedback, negative or positive, about this port is welcome
549and should be addressed to me at the e-mail addresses below.
550
551I intend creating a private mailing list for announcements of fixes &
552updates to this port. If you wish to receive such e-mail announcments,
553please send me an e-mail requesting that you be added to this list.
554
555Andrew MacIntyre
556E-mail: andymac@bullseye.apana.org.au, or andymac@pcug.org.au
557Web: http://www.andymac.org/
558
Andrew MacIntyreffcf8992002-08-18 06:28:21 +000055918 August, 2001.