5 (editors: check NEWS.help for information about editing NEWS using ReST.)
7 What's New in Python 2.5 alpha 1?
8 =================================
10 *Release date: XX-XXX-2006*
15 - PEP 352, patch #1104669: Make exceptions new-style objects. Introduced the
16 new exception base class, BaseException, which has a new message attribute.
17 KeyboardInterrupt and SystemExit to directly inherit from BaseException now.
18 Raising a string exception now raises a DeprecationWarning.
20 - Patch #1438387, PEP 328: relative and absolute imports. Imports can now be
21 explicitly relative, using 'from .module import name' to mean 'from the same
22 package as this module is in. Imports without dots still default to the
23 old relative-then-absolute, unless 'from __future__ import
24 absolute_import' is used.
26 - Properly check if 'warnings' raises an exception (usually when a filter set
27 to "error" is triggered) when raising a warning for raising string
30 - CO_GENERATOR_ALLOWED is no longer defined, this behavior is the default.
31 The name was removed from Include/code.h.
33 - PEP 308: conditional expressions were added (x if cond else y).
36 - The copy module now "copies" function objects (as atomic objects).
37 - dict.__getitem__ now looks for a __missing__ hook before raising
40 - PEP 343: with statement implemented. Needs ``from __future__ import
41 with_statement``. Use of 'with' as a variable will generate a warning.
42 Use of 'as' as a variable will also generate a warning (unless it's
43 part of an import statement).
44 The following objects have __context__ methods:
45 - The built-in file type.
46 - The thread.LockType type.
47 - The following types defined by the threading module:
48 Lock, RLock, Condition, Semaphore, BoundedSemaphore.
49 - The decimal.Context class.
51 - Fix the encodings package codec search function to only search
52 inside its own package. Fixes problem reported in patch #1433198.
54 Note: Codec packages should implement and register their own
55 codec search function. PEP 100 has the details.
57 - PEP 353: Using ssize_t as the index type.
59 - Patch #1400181, fix unicode string formatting to not use the locale.
60 This is how string objects work. u'%f' could use , instead of .
61 for the decimal point. Now both strings and unicode always use periods.
63 - Bug #1244610, #1392915, fix build problem on OpenBSD 3.7 and 3.8.
64 configure would break checking curses.h.
66 - Bug #959576: The pwd module is now builtin. This allows Python to be
67 built on UNIX platforms without $HOME set.
69 - Bug #1072182, fix some potential problems if characters are signed.
71 - Bug #889500, fix line number on SyntaxWarning for global declarations.
73 - Bug #1378022, UTF-8 files with a leading BOM crashed the interpreter.
75 - Support for converting hex strings to floats no longer works.
76 This was not portable. float('0x3') now raises a ValueError.
78 - Patch #1382163: Expose Subversion revision number to Python. New C API
79 function Py_GetBuildNumber(). New attribute sys.build_number. Build number
80 is now displayed in interactive prompt banner.
82 - Implementation of PEP 341 - Unification of try/except and try/finally.
83 "except" clauses can now be written together with a "finally" clause in
84 one try statement instead of two nested ones. Patch #1355913.
86 - Bug #1379994: Builtin unicode_escape and raw_unicode_escape codec
87 now encodes backslash correctly.
89 - Patch #1350409: Work around signal handling bug in Visual Studio 2005.
91 - Bug #1281408: Py_BuildValue now works correct even with unsigned longs
94 - SF Bug #1350188, "setdlopenflags" leads to crash upon "import"
95 It was possible dlerror() returns a NULL pointer, use a default error
98 - Replaced most Unicode charmap codecs with new ones using the
99 new Unicode translate string feature in the builtin charmap
100 codec; the codecs were created from the mapping tables available
101 at ftp.unicode.org and contain a few updates (e.g. the Mac OS
102 encodings now include a mapping for the Apple logo)
104 - Added a few more codecs for Mac OS encodings
106 - Speed up some Unicode operations.
108 - A new AST parser implementation was completed. The abstract
109 syntax tree is available for read-only (non-compile) access
110 to Python code; an _ast module was added.
112 - SF bug #1167751: fix incorrect code being for generator expressions.
113 The following code now raises a SyntaxError: foo(a = i for i in range(10))
115 - SF Bug #976608: fix SystemError when mtime of an imported file is -1.
117 - SF Bug #887946: fix segfault when redirecting stdin from a directory.
118 Provide a warning when a directory is passed on the command line.
120 - Fix segfault with invalid coding.
122 - SF bug #772896: unknown encoding results in MemoryError.
124 - All iterators now have a Boolean value of true. Formerly, some iterators
125 supported a __len__() method which evaluated to False when the iterator
128 - On 64-bit platforms, when __len__() returns a value that cannot be
129 represented as a C int, raise OverflowError.
131 - test__locale is skipped on OS X < 10.4 (only partial locale support is
134 - SF bug #893549: parsing keyword arguments was broken with a few format
137 - Changes donated by Elemental Security to make it work on AIX 5.3
138 with IBM's 64-bit compiler (SF patch #1284289). This also closes SF
139 bug #105470: test_pwd fails on 64bit system (Opteron).
141 - Changes donated by Elemental Security to make it work on HP-UX 11 on
142 Itanium2 with HP's 64-bit compiler (SF patch #1225212).
144 - Disallow keyword arguments for type constructors that don't use them
145 (fixes bug #1119418).
147 - Forward UnicodeDecodeError into SyntaxError for source encoding errors.
149 - SF bug #900092: When tracing (e.g. for hotshot), restore 'return' events for
150 exceptions that cause a function to exit.
152 - The implementation of set() and frozenset() was revised to use its
153 own internal data structure. Memory consumption is reduced by 1/3
154 and there are modest speed-ups as well. The API is unchanged.
156 - SF bug #1238681: freed pointer is used in longobject.c:long_pow().
158 - SF bug #1229429: PyObject_CallMethod failed to decrement some
159 reference counts in some error exit cases.
161 - SF bug #1185883: Python's small-object memory allocator took over
162 a block managed by the platform C library whenever a realloc specified
163 a small new size. However, there's no portable way to know then how
164 much of the address space following the pointer is valid, so no
165 portable way to copy data from the C-managed block into Python's
166 small-object space without risking a memory fault. Python's small-object
167 realloc now leaves such blocks under the control of the platform C
170 - SF bug #1232517: An overflow error was not detected properly when
171 attempting to convert a large float to an int in os.utime().
173 - SF bug #1224347: hex longs now print with lowercase letters just
174 like their int counterparts.
176 - SF bug #1163563: the original fix for bug #1010677 ("thread Module
177 Breaks PyGILState_Ensure()") broke badly in the case of multiple
178 interpreter states; back out that fix and do a better job (see
179 http://mail.python.org/pipermail/python-dev/2005-June/054258.html
180 for a longer write-up of the problem).
182 - SF patch #1180995: marshal now uses a binary format by default when
185 - SF patch #1181301: on platforms that appear to use IEEE 754 floats,
186 the routines that promise to produce IEEE 754 binary representations
187 of floats now simply copy bytes around.
189 - bug #967182: disallow opening files with 'wU' or 'aU' as specified by PEP
192 - patch #1109424: int, long, float, complex, and unicode now check for the
193 proper magic slot for type conversions when subclassed. Previously the
194 magic slot was ignored during conversion. Semantics now match the way
195 subclasses of str always behaved. int/long/float, conversion of an instance
196 to the base class has been moved to the proper nb_* magic slot and out of
198 Thanks Walter Dörwald.
200 - Descriptors defined in C with a PyGetSetDef structure, where the setter is
201 NULL, now raise an AttributeError when attempting to set or delete the
202 attribute. Previously a TypeError was raised, but this was inconsistent
203 with the equivalent pure-Python implementation.
205 - It is now safe to call PyGILState_Release() before
206 PyEval_InitThreads() (note that if there is reason to believe there
207 are multiple threads around you still must call PyEval_InitThreads()
208 before using the Python API; this fix is for extension modules that
209 have no way of knowing if Python is multi-threaded yet).
211 - Typing Ctrl-C whilst raw_input() was waiting in a build with threads
212 disabled caused a crash.
214 - Bug #1165306: instancemethod_new allowed the creation of a method
215 with im_class == im_self == NULL, which caused a crash when called.
217 - Move exception finalisation later in the shutdown process - this
218 fixes the crash seen in bug #1165761
220 - Added two new builtins, any() and all().
222 - Defining a class with empty parentheses is now allowed
223 (e.g., ``class C(): pass`` is no longer a syntax error).
224 Patch #1176012 added support to the 'parser' module and 'compiler' package
225 (thanks to logistix for that added support).
227 - Patch #1115086: Support PY_LONGLONG in structmember.
229 - Bug #1155938: new style classes did not check that __init__() was
232 - Patch #802188: Report characters after line continuation character
233 ('\') with a specific error message.
235 - Bug #723201: Raise a TypeError for passing bad objects to 'L' format.
237 - Bug #1124295: the __name__ attribute of file objects was
238 inadvertently made inaccessible in restricted mode.
240 - Bug #1074011: closing sys.std{out,err} now causes a flush() and
243 - min() and max() now support key= arguments with the same meaning as in
246 - The peephole optimizer now performs simple constant folding in expressions:
249 - set and frozenset objects can now be marshalled. SF #1098985.
251 - Bug #1077106: Poor argument checking could cause memory corruption
252 in calls to os.read().
254 - The parser did not complain about future statements in illegal
255 positions. It once again reports a syntax error if a future
256 statement occurs after anything other than a doc string.
258 - Change the %s format specifier for str objects so that it returns a
259 unicode instance if the argument is not an instance of basestring and
260 calling __str__ on the argument returns a unicode instance.
262 - Patch #1413181: changed ``PyThreadState_Delete()`` to forget about the
263 current thread state when the auto-GIL-state machinery knows about
264 it (since the thread state is being deleted, continuing to remember it
265 can't help, but can hurt if another thread happens to get created with
271 - Patch 1433928: Added a new type, defaultdict, to the collections module.
272 This uses the new __missing__ hook behavior added to dict (see above).
274 - Bug #854823: socketmodule now builds on Sun platforms even when
275 INET_ADDRSTRLEN is not defined.
277 - Patch #1393157: os.startfile() now has an optional argument to specify
278 a "command verb" to invoke on the file.
280 - Bug #876637, prevent stack corruption when socket descriptor
281 is larger than FD_SETSIZE.
283 - Patch #1407135, bug #1424041: harmonize mmap behavior of anonymous memory.
284 mmap.mmap(-1, size) now returns anonymous memory in both Unix and Windows.
285 mmap.mmap(0, size) should not be used on Windows for anonymous memory.
287 - Patch #1422385: The nis module now supports access to domains other
288 than the system default domain.
290 - Use Win32 API to implement os.stat/fstat. As a result, subsecond timestamps
291 are reported, the limit on path name lengths is removed, and stat reports
292 WindowsError now (instead of OSError).
294 - Add bsddb.db.DBEnv.set_tx_timestamp allowing time based database recovery.
296 - Bug #1413192, fix seg fault in bsddb if a transaction was deleted
299 - Patch #1103116: Basic AF_NETLINK support.
301 - Bug #1402308, (possible) segfault when using mmap.mmap(-1, ...)
303 - Bug #1400822, _curses over{lay,write} doesn't work when passing 6 ints.
304 Also fix ungetmouse() which did not accept arguments properly.
305 The code now conforms to the documented signature.
307 - Bug #1400115, Fix segfault when calling curses.panel.userptr()
308 without prior setting of the userptr.
310 - Fix 64-bit problems in bsddb.
312 - Patch #1365916: fix some unsafe 64-bit mmap methods.
314 - Bug #1290333: Added a workaround for cjkcodecs' _codecs_cn build
317 - Bug #869197: os.setgroups rejects long integer arguments
319 - Bug #1346533, select.poll() doesn't raise an error if timeout > sys.maxint
321 - Bug #1344508, Fix UNIX mmap leaking file descriptors
323 - Patch #1338314, Bug #1336623: fix tarfile so it can extract
324 REGTYPE directories from tarfiles written by old programs.
326 - Patch #1407992, fixes broken bsddb module db associate when using
327 BerkeleyDB 3.3, 4.0 or 4.1.
329 - Get bsddb module to build with BerkeleyDB version 4.4
331 - Get bsddb module to build with BerkeleyDB version 3.2
333 - Patch #1309009, Fix segfault in pyexpat when the XML document is in latin_1,
334 but Python incorrectly assumes it is in UTF-8 format
336 - Fix parse errors in the readline module when compiling without threads.
338 - Patch #1288833: Removed thread lock from socket.getaddrinfo on
339 FreeBSD 5.3 and later versions which got thread-safe getaddrinfo(3).
341 - Patches #1298449 and #1298499: Add some missing checks for error
342 returns in cStringIO.c.
344 - Patch #1297028: fix segfault if call type on MultibyteCodec,
345 MultibyteStreamReader, or MultibyteStreamWriter
347 - Fix memory leak in posix.access().
349 - Patch #1213831: Fix typo in unicodedata._getcode.
351 - Bug #1007046: os.startfile() did not accept unicode strings encoded in
352 the file system encoding.
354 - Patch #756021: Special-case socket.inet_aton('255.255.255.255') for
355 platforms that don't have inet_aton().
357 - Bug #1215928: Fix bz2.BZ2File.seek() for 64-bit file offsets.
359 - Bug #1191043: Fix bz2.BZ2File.(x)readlines for files containing one
360 line without newlines.
362 - Bug #728515: mmap.resize() now resizes the file on Unix as it did
365 - Patch #1180695: Add nanosecond stat resolution, and st_gen,
366 st_birthtime for FreeBSD.
368 - Patch #1231069: The fcntl.ioctl function now uses the 'I' code for
369 the request code argument, which results in more C-like behaviour
370 for large or negative values.
372 - Bug #1234979: For the argument of thread.Lock.acquire, the Windows
373 implementation treated all integer values except 1 as false.
375 - Bug #1194181: bz2.BZ2File didn't handle mode 'U' correctly.
377 - Patch #1212117: os.stat().st_flags is now accessible as a attribute
378 if available on the platform.
380 - Patch #1103951: Expose O_SHLOCK and O_EXLOCK in the posix module if
381 available on the platform.
383 - Bug #1166660: The readline module could segfault if hook functions
384 were set in a different thread than that which called readline.
386 - collections.deque objects now support a remove() method.
388 - operator.itemgetter() and operator.attrgetter() now support retrieving
389 multiple fields. This provides direct support for sorting on multiple
390 keys (primary, secondary, etc).
392 - os.access now supports Unicode path names on non-Win32 systems.
394 - Patches #925152, #1118602: Avoid reading after the end of the buffer
395 in pyexpat.GetInputContext.
397 - Patches #749830, #1144555: allow UNIX mmap size to default to current
400 - Added functional.partial(). See PEP309.
402 - Patch #1093585: raise a ValueError for negative history items in readline.
403 {remove_history,replace_history}
405 - The spwd module has been added, allowing access to the shadow password
408 - stat_float_times is now True.
410 - array.array objects are now picklable.
412 - the cPickle module no longer accepts the deprecated None option in the
413 args tuple returned by __reduce__().
415 - itertools.islice() now accepts None for the start and step arguments.
416 This allows islice() to work more readily with slices:
417 islice(s.start, s.stop, s.step)
419 - datetime.datetime() now has a strptime class method which can be used to
420 create datetime object using a string and format.
425 - PEP 343: new module contextlib.py defines decorator @contextmanager
426 and helpful context managers nested() and closing().
428 - The compiler package now supports future imports after the module docstring.
430 - Bug #1413790: zipfile now sanitizes absolute archive names that are
431 not allowed by the specs.
433 - Patch #1215184: FileInput now can be given an opening hook which can
434 be used to control how files are opened.
436 - Patch #1212287: fileinput.input() now has a mode parameter for
437 specifying the file mode input files should be opened with.
439 - Patch #1215184: fileinput now has a fileno() function for getting the
442 - Patch #1349274: gettext.install() now optionally installs additional
443 translation functions other than _() in the builtin namespace.
445 - Patch #1337756: fileinput now accepts Unicode filenames.
447 - Patch #1373643: The chunk module can now read chunks larger than
450 - Patch #1417555: SimpleHTTPServer now returns Last-Modified headers.
452 - Bug #1430298: It is now possible to send a mail with an empty
453 return address using smtplib.
455 - Bug #1432260: The names of lambda functions are now properly displayed
458 - Patch #1412872: zipfile now sets the creator system to 3 (Unix)
459 unless the system is Win32.
461 - Patch #1349118: urllib now supports user:pass@ style proxy
462 specifications, raises IOErrors when proxies for unsupported protocols
463 are defined, and uses the https proxy on https redirections.
465 - Bug #902075: urllib2 now supports 'host:port' style proxy specifications.
467 - Bug #1407902: Add support for sftp:// URIs to urlparse.
469 - Bug #1371247: Update Windows locale identifiers in locale.py.
471 - Bug #1394565: SimpleHTTPServer now doesn't choke on query parameters
474 - Bug #1403410: The warnings module now doesn't get confused
475 when it can't find out the module name it generates a warning for.
477 - Patch #1177307: Added a new codec utf_8_sig for UTF-8 with a BOM signature.
479 - Patch #1157027: cookielib mishandles RFC 2109 cookies in Netscape mode
481 - Patch #1117398: cookielib.LWPCookieJar and .MozillaCookieJar now raise
482 LoadError as documented, instead of IOError. For compatibility,
483 LoadError subclasses IOError.
485 - Added the hashlib module. It provides secure hash functions for MD5 and
486 SHA1, 224, 256, 384, and 512. Note that recent developments make the
487 historic MD5 and SHA1 unsuitable for cryptographic-strength applications.
488 In <http://mail.python.org/pipermail/python-dev/2005-December/058850.html>
489 Ronald L. Rivest offered this advice for Python:
491 "The consensus of researchers in this area (at least as
492 expressed at the NIST Hash Function Workshop 10/31/05),
493 is that SHA-256 is a good choice for the time being, but
494 that research should continue, and other alternatives may
495 arise from this research. The larger SHA's also seem OK."
497 - Added a subset of Fredrik Lundh's ElementTree package. Available
498 modules are xml.etree.ElementTree, xml.etree.ElementPath, and
499 xml.etree.ElementInclude, from ElementTree 1.2.6.
501 - Patch #1162825: Support non-ASCII characters in IDLE window titles.
503 - Bug #1365984: urllib now opens "data:" URLs again.
505 - Patch #1314396: prevent deadlock for threading.Thread.join() when an exception
506 is raised within the method itself on a previous call (e.g., passing in an
509 - Bug #1340337: change time.strptime() to always return ValueError when there
510 is an error in the format string.
512 - Patch #754022: Greatly enhanced webbrowser.py (by Oleg Broytmann).
514 - Bug #729103: pydoc.py: Fix docother() method to accept additional
517 - Patch #1300515: xdrlib.py: Fix pack_fstring() to really use null bytes
520 - Bug #1296004: httplib.py: Limit maximal amount of data read from the
521 socket to avoid a MemoryError on Windows.
523 - Patch #1166948: locale.py: Prefer LC_ALL, LC_CTYPE and LANG over LANGUAGE
524 to get the correct encoding.
526 - Patch #1166938: locale.py: Parse LANGUAGE as a colon separated list of
529 - Patch #1268314: Cache lines in StreamReader.readlines for performance.
531 - Bug #1290505: Fix clearing the regex cache for time.strptime().
533 - Bug #1167128: Fix size of a symlink in a tarfile to be 0.
535 - Patch #810023: Fix off-by-one bug in urllib.urlretrieve reporthook
538 - Bug #1163178: Make IDNA return an empty string when the input is empty.
540 - Patch #848017: Make Cookie more RFC-compliant. Use CRLF as default output
541 separator and do not output trailing semicola.
543 - Patch #1062060: urllib.urlretrieve() now raises a new exception, named
544 ContentTooShortException, when the actually downloaded size does not
545 match the Content-Length header.
547 - Bug #1121494: distutils.dir_utils.mkpath now accepts Unicode strings.
549 - Bug #1178484: Return complete lines from codec stream readers
550 even if there is an exception in later lines, resulting in
551 correct line numbers for decoding errors in source code.
553 - Bug #1192315: Disallow negative arguments to clear() in pdb.
555 - Patch #827386: Support absolute source paths in msvccompiler.py.
557 - Patch #1105730: Apply the new implementation of commonprefix in posixpath
558 to ntpath, macpath, os2emxpath and riscospath.
560 - Fix a problem in Tkinter introduced by SF patch #869468: delete bogus
561 __hasattr__ and __delattr__ methods on class Tk that were breaking
564 - Bug #1015140: disambiguated the term "article id" in nntplib docs and
565 docstrings to either "article number" or "message id".
567 - Bug #1238170: threading.Thread.__init__ no longer has "kwargs={}" as a
568 parameter, but uses the usual "kwargs=None".
570 - textwrap now processes text chunks at O(n) speed instead of O(n**2).
571 Patch #1209527 (Contributed by Connelly).
573 - urllib2 has now an attribute 'httpresponses' mapping from HTTP status code
574 to W3C name (404 -> 'Not Found'). RFE #1216944.
576 - Bug #1177468: Don't cache the /dev/urandom file descriptor for os.urandom,
577 as this can cause problems with apps closing all file descriptors.
579 - Bug #839151: Fix an attempt to access sys.argv in the warnings module
580 it can be missing in embedded interpreters
582 - Bug #1155638: Fix a bug which affected HTTP 0.9 responses in httplib.
584 - Bug #1100201: Cross-site scripting was possible on BaseHTTPServer via
587 - Bug #1108948: Cookie.py produced invalid JavaScript code.
589 - The tokenize module now detects and reports indentation errors.
592 - The tokenize module has a new untokenize() function to support a full
593 roundtrip from lexed tokens back to Python sourcecode. In addition,
594 the generate_tokens() function now accepts a callable argument that
595 terminates by raising StopIteration.
597 - Bug #1196315: fix weakref.WeakValueDictionary constructor.
599 - Bug #1213894: os.path.realpath didn't resolve symlinks that were the first
600 component of the path.
602 - Patch #1120353: The xmlrpclib module provides better, more transparent,
603 support for datetime.{datetime,date,time} objects. With use_datetime set
604 to True, applications shouldn't have to fiddle with the DateTime wrapper
607 - distutils.commands.upload was added to support uploading distribution
610 - distutils.commands.register now encodes the data as UTF-8 before posting
613 - decimal operator and comparison methods now return NotImplemented
614 instead of raising a TypeError when interacting with other types. This
615 allows other classes to implement __radd__ style methods and have them
618 - Bug #1163325: Decimal infinities failed to hash. Attempting to
619 hash a NaN raised an InvalidOperation instead of a TypeError.
621 - Patch #918101: Add tarfile open mode r|* for auto-detection of the
622 stream compression; add, for symmetry reasons, r:* as a synonym of r.
624 - Patch #1043890: Add extractall method to tarfile.
626 - Patch #1075887: Don't require MSVC in distutils if there is nothing
629 - Patch #1103407: Properly deal with tarfile iterators when untarring
630 symbolic links on Windows.
632 - Patch #645894: Use getrusage for computing the time consumption in
633 profile.py if available.
635 - Patch #1046831: Use get_python_version where appropriate in sysconfig.py.
637 - Patch #1117454: Remove code to special-case cookies without values
640 - Patch #1117339: Add cookielib special name tests.
642 - Patch #1112812: Make bsddb/__init__.py more friendly for modulefinder.
644 - Patch #1110248: SYNC_FLUSH the zlib buffer for GZipFile.flush.
646 - Patch #1107973: Allow to iterate over the lines of a tarfile.ExFileObject.
648 - Patch #1104111: Alter setup.py --help and --help-commands.
650 - Patch #1121234: Properly cleanup _exit and tkerror commands.
652 - Patch #1049151: xdrlib now unpacks booleans as True or False.
654 - Fixed bug in a NameError bug in cookielib. Patch #1116583.
656 - Applied a security fix to SimpleXMLRPCserver (PSF-2005-001). This
657 disables recursive traversal through instance attributes, which can
658 be exploited in various ways.
660 - Bug #1222790: in SimpleXMLRPCServer, set the reuse-address and close-on-exec
661 flags on the HTTP listening socket.
663 - Bug #792570: SimpleXMLRPCServer had problems if the request grew too large.
664 Fixed by reading the HTTP body in chunks instead of one big socket.read().
666 - Patches #893642, #1039083: add allow_none, encoding arguments to constructors of
667 SimpleXMLRPCServer and CGIXMLRPCRequestHandler.
669 - Bug #1110478: Revert os.environ.update to do putenv again.
671 - Bug #1103844: fix distutils.install.dump_dirs() with negated options.
673 - os.{SEEK_SET, SEEK_CUR, SEEK_END} have been added for convenience.
675 - Enhancements to the csv module:
677 + Dialects are now validated by the underlying C code, better
678 reflecting its capabilities, and improving its compliance with
680 + Dialect parameter parsing has been re-implemented to improve error
682 + quotechar=None and quoting=QUOTE_NONE now work the way PEP 305
684 + the parser now removes the escapechar prefix from escaped characters.
685 + when quoting=QUOTE_NONNUMERIC, the writer now tests for numeric
686 types, rather than any object than can be represented as a numeric.
687 + when quoting=QUOTE_NONNUMERIC, the reader now casts unquoted fields
689 + reader now allows \r characters to be quoted (previously it only allowed
691 + writer doublequote handling improved.
692 + Dialect classes passed to the module are no longer instantiated by
693 the module before being parsed (the former validation scheme required
694 this, but the mechanism was unreliable).
695 + The dialect registry now contains instances of the internal
696 C-coded dialect type, rather than references to python objects.
697 + the internal c-coded dialect type is now immutable.
698 + register_dialect now accepts the same keyword dialect specifications
699 as the reader and writer, allowing the user to register dialects
700 without first creating a dialect class.
701 + a configurable limit to the size of parsed fields has been added -
702 previously, an unmatched quote character could result in the entire
703 file being read into the field buffer before an error was reported.
704 + A new module method csv.field_size_limit() has been added that sets
705 the parser field size limit (returning the former limit). The initial
707 + A line_num attribute has been added to the reader object, which tracks
708 the number of lines read from the source iterator. This is not
709 the same as the number of records returned, as records can span
711 + reader and writer objects were not being registered with the cyclic-GC.
714 - _DummyThread objects in the threading module now delete self.__block that is
715 inherited from _Thread since it uses up a lock allocated by 'thread'. The
716 lock primitives tend to be limited in number and thus should not be wasted on
717 a _DummyThread object. Fixes bug #1089632.
719 - The imghdr module now detects Exif files.
721 - StringIO.truncate() now correctly adjusts the size attribute.
724 - locale.py now uses an updated locale alias table (built using
725 Tools/i18n/makelocalealias.py, a tool to parse the X11 locale
726 alias file); the encoding lookup was enhanced to use Python's
727 encoding alias table.
729 - moved deprecated modules to Lib/lib-old: whrandom, tzparse, statcache.
731 - the pickle module no longer accepts the deprecated None option in the
732 args tuple returned by __reduce__().
734 - optparse now optionally imports gettext. This allows its use in setup.py.
736 - the pickle module no longer uses the deprecated bin parameter.
738 - the shelve module no longer uses the deprecated binary parameter.
740 - the pstats module no longer uses the deprecated ignore() method.
742 - the filecmp module no longer uses the deprecated use_statcache argument.
744 - unittest.TestCase.run() and unittest.TestSuite.run() can now be successfully
745 extended or overridden by subclasses. Formerly, the subclassed method would
746 be ignored by the rest of the module. (Bug #1078905).
748 - heapq.nsmallest() and heapq.nlargest() now support key= arguments with
749 the same meaning as in list.sort().
751 - Bug #1076985: ``codecs.StreamReader.readline()`` now calls ``read()`` only
752 once when a size argument is given. This prevents a buffer overflow in the
753 tokenizer with very long source lines.
755 - Bug #1083110: ``zlib.decompress.flush()`` would segfault if called
756 immediately after creating the object, without any intervening
757 ``.decompress()`` calls.
759 - The reconvert.quote function can now emit triple-quoted strings. The
760 reconvert module now has some simple documentation.
762 - ``UserString.MutableString`` now supports negative indices in
763 ``__setitem__`` and ``__delitem__``
765 - Bug #1149508: ``textwrap`` now handles hyphenated numbers (eg. "2004-03-05")
768 - Partial fixes for SF bugs #1163244 and #1175396: If a chunk read by
769 ``codecs.StreamReader.readline()`` has a trailing "\r", read one more
770 character even if the user has passed a size parameter to get a proper
771 line ending. Remove the special handling of a "\r\n" that has been split
774 - Bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
775 about illegal code points. The codec now supports PEP 293 style error
778 - Bug #1235646: ``codecs.StreamRecoder.next()`` now reencodes the data it reads
779 from the input stream, so that the output is a byte string in the correct
780 encoding instead of a unicode string.
782 - Bug #1202493: Fixing SRE parser to handle '{}' as perl does, rather than
783 considering it exactly like a '*'.
785 - Bug #1245379: Add "unicode-1-1-utf-7" as an alias for "utf-7" to
786 ``encodings.aliases``.
788 - ` uu.encode()`` and ``uu.decode()`` now support unicode filenames.
790 - Patch #1413711: Certain patterns of differences were making difflib
791 touch the recursion limit.
796 - Patch #1432345: Make python compile on DragonFly.
798 - Build support for Win64-AMD64 was added.
800 - Patch #1428494: Prefer linking against ncursesw over ncurses library.
802 - Patch #881820: look for openpty and forkpty also in libbsd.
804 - The sources of zlib are now part of the Python distribution (zlib 1.2.3).
805 The zlib module is now builtin on Windows.
807 - Use -xcode=pic32 for CCSHARED on Solaris with SunPro.
809 - Bug #1189330: configure did not correctly determine the necessary
810 value of LINKCC if python was built with GCC 4.0.
812 - Upgrade Windows build to zlib 1.2.3 which eliminates a potential security
813 vulnerability in zlib 1.2.1 and 1.2.2.
815 - EXTRA_CFLAGS has been introduced as an environment variable to hold compiler
816 flags that change binary compatibility. Changes were also made to
817 distutils.sysconfig to also use the environment variable when used during
818 compilation of the interpreter and of C extensions through distutils.
820 - SF patch 1171735: Darwin 8's headers are anal about POSIX compliance,
821 and linking has changed (prebinding is now deprecated, and libcc_dynamic
822 no longer exists). This configure patch makes things right.
824 - Bug #1158607: Build with --disable-unicode again.
826 - spwdmodule.c is built only if either HAVE_GETSPNAM or HAVE_HAVE_GETSPENT is
827 defined. Discovered as a result of not being able to build on OS X.
829 - setup.py now uses the directories specified in LDFLAGS using the -L option
830 and in CPPFLAGS using the -I option for adding library and include
831 directories, respectively, for compiling extension modules against. This has
832 led to the core being compiled using the values in CPPFLAGS. It also removes
833 the need for the special-casing of both DarwinPorts and Fink for darwin since
834 the proper directories can be specified in LDFLAGS (``-L/sw/lib`` for Fink,
835 ``-L/opt/local/lib`` for DarwinPorts) and CPPFLAGS (``-I/sw/include`` for
836 Fink, ``-I/opt/local/include`` for DarwinPorts).
838 - Test in configure.in that checks for tzset no longer dependent on tm->tm_zone
839 to exist in the struct (not required by either ISO C nor the UNIX 2 spec).
840 Tests for sanity in tzname when HAVE_TZNAME defined were also defined.
841 Closes bug #1096244. Thanks Gregory Bond.
846 - Added a C API for set and frozenset objects.
848 - Removed PyRange_New().
850 - Patch #1313939: PyUnicode_DecodeCharmap() accepts a unicode string as the
851 mapping argument now. This string is used as a mapping table. Byte values
852 greater than the length of the string and 0xFFFE are treated as undefined
859 - In test_os, st_?time is now truncated before comparing it with ST_?TIME.
861 - Patch #1276356: New resource "urlfetch" is implemented. This enables
862 even impatient people to run tests that require remote files.
868 - Bug #1402224: Add warning to dl docs about crashes.
870 - Bug #1396471: Document that Windows' ftell() can return invalid
871 values for text files with UNIX-style line endings.
873 - Bug #1274828: Document os.path.splitunc().
875 - Bug #1190204: Clarify which directories are searched by site.py.
877 - Bug #1193849: Clarify os.path.expanduser() documentation.
879 - Bug #1243192: re.UNICODE and re.LOCALE affect \d, \D, \s and \S.
881 - Bug #755617: Document the effects of os.chown() on Windows.
883 - Patch #1180012: The documentation for modulefinder is now in the library reference.
885 - Patch #1213031: Document that os.chown() accepts argument values of -1.
887 - Bug #1190563: Document os.waitpid() return value with WNOHANG flag.
889 - Bug #1175022: Correct the example code for property().
891 - Document the IterableUserDict class in the UserDict module.
894 - Remove all latent references for "Macintosh" that referred to semantics for
895 Mac OS 9 and change to reflect the state for OS X.
896 Closes patch #1095802. Thanks Jack Jansen.
905 - FreeBSD 7 support is added.
911 - Created Misc/Vim/vim_syntax.py to auto-generate a python.vim file in that
912 directory for syntax highlighting in Vim. Vim directory was added and placed
913 vimrc to it (was previous up a level).
915 - Added two new files to Tools/scripts: pysource.py, which recursively
916 finds Python source files, and findnocoding.py, which finds Python
917 source files that need an encoding declaration.
918 Patch #784089, credits to Oleg Broytmann.
920 - Bug #1072853: pindent.py used an uninitialized variable.
922 - Patch #1177597: Correct Complex.__init__.
924 - Fixed a display glitch in Pynche, which could cause the right arrow to
925 wiggle over by a pixel.
927 What's New in Python 2.4 final?
928 ===============================
930 *Release date: 30-NOV-2004*
935 - Bug 875692: Improve signal handling, especially when using threads, by
936 forcing an early re-execution of PyEval_EvalFrame() "periodic" code when
937 things_to_do is not cleared by Py_MakePendingCalls().
940 What's New in Python 2.4 (release candidate 1)
941 ==============================================
943 *Release date: 18-NOV-2004*
948 - Bug 1061968: Fixes in 2.4a3 to address thread bug 1010677 reintroduced
949 the years-old thread shutdown race bug 225673. Numeric history lesson
950 aside, all bugs in all three reports are fixed now.
956 - Bug 1052242: If exceptions are raised by an atexit handler function an
957 attempt is made to execute the remaining handlers. The last exception
960 - ``doctest``'s new support for adding ``pdb.set_trace()`` calls to
961 doctests was broken in a dramatic but shallow way. Fixed.
963 - Bug 1065388: ``calendar``'s ``day_name``, ``day_abbr``, ``month_name``,
964 and ``month_abbr`` attributes emulate sequences of locale-correct
965 spellings of month and day names. Because the locale can change at
966 any time, the correct spelling is recomputed whenever one of these is
967 indexed. In the worst case, the index may be a slice object, so these
968 recomputed every day or month name each time they were indexed. This is
969 much slower than necessary in the usual case, when the index is just an
970 integer. In that case, only the single spelling needed is recomputed
971 now; and, when the index is a slice object, only the spellings needed
972 by the slice are recomputed now.
974 - Patch 1061679: Added ``__all__`` to pickletools.py.
979 - Bug 1034277 / Patch 1035255: Remove compilation of core against CoreServices
980 and CoreFoundation on OS X. Involved removing PyMac_GetAppletScriptFile()
981 which has no known users. Thanks Bob Ippolito.
986 - The PyRange_New() function is deprecated.
989 What's New in Python 2.4 beta 2?
990 ================================
992 *Release date: 03-NOV-2004*
997 The Python Software Foundation changed the license under which Python
998 is released, to remove Python version numbers. There were no other
999 changes to the license. So, for example, wherever the license for
1000 Python 2.3 said "Python 2.3", the new license says "Python". The
1001 intent is to make it possible to refer to the PSF license in a more
1002 durable way. For example, some people say they're confused by that
1003 the Open Source Initiative's entry for the Python Software Foundation
1006 http://www.opensource.org/licenses/PythonSoftFoundation.php
1008 says "Python 2.1.1" all over it, wondering whether it applies only
1011 The official name of the new license is the Python Software Foundation
1017 - Bug #1055820 Cyclic garbage collection was not protecting against that
1018 calling a live weakref to a piece of cyclic trash could resurrect an
1019 insane mutation of the trash if any Python code ran during gc (via
1020 running a dead object's __del__ method, running another callback on a
1021 weakref to a dead object, or via any Python code run in any other thread
1022 that managed to obtain the GIL while a __del__ or callback was running
1023 in the thread doing gc). The most likely symptom was "impossible"
1024 ``AttributeError`` exceptions, appearing seemingly at random, on weakly
1025 referenced objects. The cure was to clear all weakrefs to unreachable
1026 objects before allowing any callbacks to run.
1028 - Bug #1054139 _PyString_Resize() now invalidates its cached hash value.
1033 - Bug #1048870: the compiler now generates distinct code objects for
1034 functions with identical bodies. This was producing confusing
1035 traceback messages which pointed to the function where the code
1036 object was first defined rather than the function being executed.
1041 - Patch #1056967 changes the semantics of Template.safe_substitute() so that
1042 no ValueError is raised on an 'invalid' match group. Now the delimiter is
1045 - Bug #1052503 pdb.runcall() was not passing along keyword arguments.
1047 - Bug #902037: XML.sax.saxutils.prepare_input_source() now combines relative
1048 paths with a base path before checking os.path.isfile().
1050 - The whichdb module can now be run from the command line.
1052 - Bug #1045381: time.strptime() can now infer the date using %U or %W (week of
1053 the year) when the day of the week and year are also specified.
1055 - Bug #1048816: fix bug in Ctrl-K at start of line in curses.textpad.Textbox
1057 - Bug #1017553: fix bug in tarfile.filemode()
1059 - Patch #737473: fix bug that old source code is shown in tracebacks even if
1060 the source code is updated and reloaded.
1065 - Patch #1044395: --enable-shared is allowed in FreeBSD also.
1067 What's New in Python 2.4 beta 1?
1068 ================================
1070 *Release date: 15-OCT-2004*
1075 - Patch #975056: Restartable signals were not correctly disabled on
1076 BSD systems. Consistently use PyOS_setsig() instead of signal().
1078 - The internal portable implementation of thread-local storage (TLS), used
1079 by the ``PyGILState_Ensure()``/``PyGILState_Release()`` API, was not
1080 thread-correct. This could lead to a variety of problems, up to and
1081 including segfaults. See bug 1041645 for an example.
1083 - Added a command line option, -m module, which searches sys.path for the
1084 module and then runs it. (Contributed by Nick Coghlan.)
1086 - The bytecode optimizer now folds tuples of constants into a single
1089 - SF bug #513866: Float/long comparison anomaly. Prior to 2.4b1, when
1090 an integer was compared to a float, the integer was coerced to a float.
1091 That could yield spurious overflow errors (if the integer was very
1092 large), and to anomalies such as
1093 ``long(1e200)+1 == 1e200 == long(1e200)-1``. Coercion to float is no
1094 longer performed, and cases like ``long(1e200)-1 < 1e200``,
1095 ``long(1e200)+1 > 1e200`` and ``(1 << 20000) > 1e200`` are computed
1101 - ``collections.deque`` objects didn't play quite right with garbage
1102 collection, which could lead to a segfault in a release build, or
1103 an assert failure in a debug build. Also, added overflow checks,
1104 better detection of mutation during iteration, and shielded deque
1105 comparisons from unusual subclass overrides of the __iter__() method.
1110 - Patch 1046644: distutils build_ext grew two new options - --swig for
1111 specifying the swig executable to use, and --swig-opts to specify
1112 options to pass to swig. --swig-opts="-c++" is the new way to spell
1115 - Patch 983206: distutils now obeys environment variable LDSHARED, if
1118 - Added Peter Astrand's subprocess.py module. See PEP 324 for details.
1120 - time.strptime() now properly escapes timezones and all other locale-specific
1121 strings for regex-specific symbols. Was breaking under Japanese Windows when
1122 the timezone was specified as "Tokyo (standard time)".
1123 Closes bug #1039270.
1125 - Updates for the email package:
1127 + email.Utils.formatdate() grew a 'usegmt' argument for HTTP support.
1128 + All deprecated APIs that in email 2.x issued warnings have been removed:
1129 _encoder argument to the MIMEText constructor, Message.add_payload(),
1130 Utils.dump_address_pair(), Utils.decode(), Utils.encode()
1131 + New deprecations: Generator.__call__(), Message.get_type(),
1132 Message.get_main_type(), Message.get_subtype(), the 'strict' argument to
1133 the Parser constructor. These will be removed in email 3.1.
1134 + Support for Python earlier than 2.3 has been removed (see PEP 291).
1135 + All defect classes have been renamed to end in 'Defect'.
1136 + Some FeedParser fixes; also a MultipartInvariantViolationDefect will be
1137 added to messages that claim to be multipart but really aren't.
1138 + Updates to documentation.
1140 - re's findall() and finditer() functions now take an optional flags argument
1141 just like the compile(), search(), and match() functions. Also, documented
1142 the previously existing start and stop parameters for the findall() and
1143 finditer() methods of regular expression objects.
1145 - rfc822 Messages now support iterating over the headers.
1147 - The (undocumented) tarfile.Tarfile.membernames has been removed;
1148 applications should use the getmember function.
1150 - httplib now offers symbolic constants for the HTTP status codes.
1152 - SF bug #1028306: Trying to compare a ``datetime.date`` to a
1153 ``datetime.datetime`` mistakenly compared only the year, month and day.
1154 Now it acts like a mixed-type comparison: ``False`` for ``==``,
1155 ``True`` for ``!=``, and raises ``TypeError`` for other comparison
1156 operators. Because datetime is a subclass of date, comparing only the
1157 base class (date) members can still be done, if that's desired, by
1158 forcing using of the approprate date method; e.g.,
1159 ``a_date.__eq__(a_datetime)`` is true if and only if the year, month
1160 and day members of ``a_date`` and ``a_datetime`` are equal.
1162 - bdist_rpm now supports command line options --force-arch,
1163 {pre,post}-install, {pre,post}-uninstall, and
1164 {prep,build,install,clean,verify}-script.
1166 - SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
1167 decoding incomplete input (when the input stream is temporarily exhausted).
1168 ``codecs.StreamReader`` now implements buffering, which enables proper
1169 readline support for the UTF-16 decoders. ``codecs.StreamReader.read()``
1170 has a new argument ``chars`` which specifies the number of characters to
1171 return. ``codecs.StreamReader.readline()`` and
1172 ``codecs.StreamReader.readlines()`` have a new argument ``keepends``.
1173 Trailing "\n"s will be stripped from the lines if ``keepends`` is false.
1175 - The documentation for doctest is greatly expanded, and now covers all
1176 the new public features (of which there are many).
1178 - ``doctest.master`` was put back in, and ``doctest.testmod()`` once again
1179 updates it. This isn't good, because every ``testmod()`` call
1180 contributes to bloating the "hidden" state of ``doctest.master``, but
1181 some old code apparently relies on it. For now, all we can do is
1182 encourage people to stitch doctests together via doctest's unittest
1183 integration features instead.
1185 - httplib now handles ipv6 address/port pairs.
1187 - SF bug #1017864: ConfigParser now correctly handles default keys,
1188 processing them with ``ConfigParser.optionxform`` when supplied,
1189 consistent with the handling of config file entries and runtime-set
1192 - SF bug #997050: Document, test, & check for non-string values in
1193 ConfigParser. Moved the new string-only restriction added in
1194 rev. 1.65 to the SafeConfigParser class, leaving existing
1195 ConfigParser & RawConfigParser behavior alone, and documented the
1196 conditions under which non-string values work.
1201 - Building on darwin now includes /opt/local/include and /opt/local/lib for
1202 building extension modules. This is so as to include software installed as
1203 a DarwinPorts port <http://darwinports.opendarwin.org/>
1205 - pyport.h now defines a Py_IS_NAN macro. It works as-is when the
1206 platform C computes true for ``x != x`` if and only if X is a NaN.
1207 Other platforms can override the default definition with a platform-
1208 specific spelling in that platform's pyconfig.h. You can also override
1209 pyport.h's default Py_IS_INFINITY definition now.
1214 - SF patch 1044089: New function ``PyEval_ThreadsInitialized()`` returns
1215 non-zero if PyEval_InitThreads() has been called.
1217 - The undocumented and unused extern int ``_PyThread_Started`` was removed.
1219 - The C API calls ``PyInterpreterState_New()`` and ``PyThreadState_New()``
1220 are two of the very few advertised as being safe to call without holding
1221 the GIL. However, this wasn't true in a debug build, as bug 1041645
1222 demonstrated. In a debug build, Python redirects the ``PyMem`` family
1223 of calls to Python's small-object allocator, to get the benefit of
1224 its extra debugging capabilities. But Python's small-object allocator
1225 isn't threadsafe, relying on the GIL to avoid the expense of doing its
1226 own locking. ``PyInterpreterState_New()`` and ``PyThreadState_New()``
1227 call the platform ``malloc()`` directly now, regardless of build type.
1229 - PyLong_AsUnsignedLong[Mask] now support int objects as well.
1231 - SF patch #998993: ``PyUnicode_DecodeUTF8Stateful`` and
1232 ``PyUnicode_DecodeUTF16Stateful`` have been added, which implement stateful
1238 - test__locale ported to unittest
1243 - ``plistlib`` now supports non-dict root objects. There is also a new
1244 interface for reading and writing plist files: ``readPlist(pathOrFile)``
1245 and ``writePlist(rootObject, pathOrFile)``
1250 - The text file comparison scripts ``ndiff.py`` and ``diff.py`` now
1251 read the input files in universal-newline mode. This spares them
1252 from consuming a great deal of time to deduce the useless result that,
1253 e.g., a file with Windows line ends and a file with Linux line ends
1254 have no lines in common.
1257 What's New in Python 2.4 alpha 3?
1258 =================================
1260 *Release date: 02-SEP-2004*
1265 - SF patch #1007189: ``from ... import ...`` statements now allow the name
1266 list to be surrounded by parentheses.
1268 - Some speedups for long arithmetic, thanks to Trevor Perrin. Gradeschool
1269 multiplication was sped a little by optimizing the C code. Gradeschool
1270 squaring was sped by about a factor of 2, by exploiting that about half
1271 the digit products are duplicates in a square. Because exponentiation
1272 uses squaring often, this also speeds long power. For example, the time
1273 to compute 17**1000000 dropped from about 14 seconds to 9 on my box due
1274 to this much. The cutoff for Karatsuba multiplication was raised,
1275 since gradeschool multiplication got quicker, and the cutoff was
1276 aggressively small regardless. The exponentiation algorithm was switched
1277 from right-to-left to left-to-right, which is more efficient for small
1278 bases. In addition, if the exponent is large, the algorithm now does
1279 5 bits (instead of 1 bit) at a time. That cut the time to compute
1280 17**1000000 on my box in half again, down to about 4.5 seconds.
1282 - OverflowWarning is no longer generated. PEP 237 scheduled this to
1283 occur in Python 2.3, but since OverflowWarning was disabled by default,
1284 nobody realized it was still being generated. On the chance that user
1285 code is still using them, the Python builtin OverflowWarning, and
1286 corresponding C API PyExc_OverflowWarning, will exist until Python 2.5.
1288 - Py_InitializeEx has been added.
1290 - Fix the order of application of decorators. The proper order is bottom-up;
1291 the first decorator listed is the last one called.
1293 - SF patch #1005778. Fix a seg fault if the list size changed while
1294 calling list.index(). This could happen if a rich comparison function
1297 - The ``func_name`` (a.k.a. ``__name__``) attribute of user-defined
1298 functions is now writable.
1300 - code_new (a.k.a new.code()) now checks its arguments sufficiently
1301 carefully that passing them on to PyCode_New() won't trigger calls
1302 to Py_FatalError() or PyErr_BadInternalCall(). It is still the case
1303 that the returned code object might be entirely insane.
1305 - Subclasses of string can no longer be interned. The semantics of
1306 interning were not clear here -- a subclass could be mutable, for
1307 example -- and had bugs. Explicitly interning a subclass of string
1308 via intern() will raise a TypeError. Internal operations that attempt
1309 to intern a string subclass will have no effect.
1311 - Bug 1003935: xrange() could report bogus OverflowErrors. Documented
1312 what xrange() intends, and repaired tests accordingly.
1317 - difflib now supports HTML side-by-side diff.
1319 - os.urandom has been added for systems that support sources of random
1322 - Patch 1012740: truncate() on a writeable cStringIO now resets the
1323 position to the end of the stream. This is consistent with the original
1324 StringIO module and avoids inadvertently resurrecting data that was
1325 supposed to have been truncated away.
1327 - Added socket.socketpair().
1329 - Added CurrentByteIndex, CurrentColumnNumber, CurrentLineNumber
1330 members to xml.parsers.expat.XMLParser object.
1332 - The mpz, rotor, and xreadlines modules, all deprecated in earlier
1333 versions of Python, have now been removed.
1338 - Patch #934356: if a module defines __all__, believe that rather than using
1339 heuristics for filtering out imported names.
1341 - Patch #941486: added os.path.lexists(), which returns True for broken
1342 symlinks, unlike os.path.exists().
1344 - the random module now uses os.urandom() for seeding if it is available.
1345 Added a new generator based on os.urandom().
1347 - difflib and diff.py can now generate HTML.
1349 - bdist_rpm now includes version and release in the BuildRoot, and
1350 replaces - by ``_`` in version and release.
1352 - distutils build/build_scripts now has an -e option to specify the
1353 path to the Python interpreter for installed scripts.
1355 - PEP 292 classes Template and SafeTemplate are added to the string module.
1357 - tarfile now generates GNU tar files by default.
1359 - HTTPResponse has now a getheaders method.
1361 - Patch #1006219: let inspect.getsource handle '@' decorators. Thanks Simon
1364 - logging.handlers.SMTPHandler.date_time has been removed;
1365 the class now uses email.Utils.formatdate to generate the time stamp.
1367 - A new function tkFont.nametofont was added to return an existing
1368 font. The Font class constructor now has an additional exists argument
1369 which, if True, requests to return/configure an existing font, rather
1370 than creating a new one.
1372 - Updated the decimal package's min() and max() methods to match the
1373 latest revision of the General Decimal Arithmetic Specification.
1374 Quiet NaNs are ignored and equal values are sorted based on sign
1377 - The decimal package's Context.copy() method now returns deep copies.
1379 - Deprecated sys.exitfunc in favor of the atexit module. The sys.exitfunc
1380 attribute will be kept around for backwards compatibility and atexit
1381 will just become the one preferred way to do it.
1383 - patch #675551: Add get_history_item and replace_history_item functions
1384 to the readline module.
1386 - bug #989672: pdb.doc and the help messages for the help_d and help_u methods
1387 of the pdb.Pdb class gives have been corrected. d(own) goes to a newer
1388 frame, u(p) to an older frame, not the other way around.
1390 - bug #990669: os.path.realpath() will resolve symlinks before normalizing the
1391 path, as normalizing the path may alter the meaning of the path if it
1394 - bug #851123: shutil.copyfile will raise an exception when trying to copy a
1395 file onto a link to itself. Thanks Gregory Ball.
1397 - bug #570300: Fix inspect to resolve file locations using os.path.realpath()
1398 so as to properly list all functions in a module when the module itself is
1399 reached through a symlink. Thanks Johannes Gijsbers.
1401 - doctest refactoring continued. See the docs for details. As part of
1402 this effort, some old and little- (never?) used features are now
1403 deprecated: the Tester class, the module is_private() function, and the
1404 isprivate argument to testmod(). The Tester class supplied a feeble
1405 "by hand" way to combine multiple doctests, if you knew exactly what
1406 you were doing. The newer doctest features for unittest integration
1407 already did a better job of that, are stronger now than ever, and the
1408 new DocTestRunner class is a saner foundation if you want to do it by
1409 hand. The "private name" filtering gimmick was a mistake from the
1410 start, and testmod() changed long ago to ignore it by default. If
1411 you want to filter out tests, the new DocTestFinder class can be used
1412 to return a list of all doctests, and you can filter that list by
1413 any computable criteria before passing it to a DocTestRunner instance.
1415 - Bug #891637, patch #1005466: fix inspect.getargs() crash on def foo((bar)).
1420 - IDLE's shortcut keys for windows are now case insensitive so that
1421 Control-V works the same as Control-v.
1423 - pygettext.py: Generate POT-Creation-Date header in ISO format.
1428 - Backward incompatibility: longintrepr.h now triggers a compile-time
1429 error if SHIFT (the number of bits in a Python long "digit") isn't
1430 divisible by 5. This new requirement allows simple code for the new
1431 5-bits-at-a-time long_pow() implementation. If necessary, the
1432 restriction could be removed (by complicating long_pow(), or by
1433 falling back to the 1-bit-at-a-time algorithm), but there are no
1436 - bug #991962: When building with --disable-toolbox-glue on Darwin no
1437 attempt to build Mac-specific modules occurs.
1439 - The --with-tsc flag to configure to enable VM profiling with the
1440 processor's timestamp counter now works on PPC platforms.
1442 - patch #1006629: Define _XOPEN_SOURCE to 500 on Solaris 8/9 to match
1443 GCC's definition and avoid redefinition warnings.
1445 - Detect pthreads support (provided by gnu pth pthread emulation) on
1448 - bug #1005737, #1007249: Fixed several build problems and warnings
1449 found on old/legacy C compilers of HP-UX, IRIX and Tru64.
1459 - patch #1005936, bug #1009373: fix index entries which contain
1460 an underscore when viewed with Acrobat.
1462 - bug #990669: os.path.normpath may alter the meaning of a path if
1463 it contains symbolic links. This has been documented in a comment
1464 since 1992, but is now in the library reference as well.
1469 - FreeBSD 6 is now supported.
1479 - Boosted the stack reservation for python.exe and pythonw.exe from
1480 the default 1MB to 2MB. Stack frames under VC 7.1 for 2.4 are enough
1481 bigger than under VC 6.0 for 2.3.4 that deeply recursive progams
1482 within the default sys.getrecursionlimit() default value of 1000 were
1483 able to suffer undetected C stack overflows. The standard test program
1484 test_compiler was one such program. If a Python process on Windows
1485 "just vanishes" without a trace, and without an error message of any
1486 kind, but with an exit code of 128, undetected stack overflow may be
1495 What's New in Python 2.4 alpha 2?
1496 =================================
1498 *Release date: 05-AUG-2004*
1503 - Patch #980695: Implements efficient string concatenation for statements
1504 of the form s=s+t and s+=t. This will vary across implementations.
1505 Accordingly, the str.join() method is strongly preferred for performance
1508 - PEP-0318, Function Decorators have been added to the language. These are
1509 implemented using the Java-style @decorator syntax, like so::
1514 (The PEP needs to be updated to reflect the current state)
1516 - When importing a module M raises an exception, Python no longer leaves M
1517 in sys.modules. Before 2.4a2 it did, and a subsequent import of M would
1518 succeed, picking up a module object from sys.modules reflecting as much
1519 of the initialization of M as completed before the exception was raised.
1520 Subsequent imports got no indication that M was in a partially-
1521 initialized state, and the importers could get into arbitrarily bad
1522 trouble as a result (the M they got was in an unintended state,
1523 arbitrarily far removed from M's author's intent). Now subsequent
1524 imports of M will continue raising exceptions (but if, for example, the
1525 source code for M is edited between import attempts, then perhaps later
1526 attempts will succeed, or raise a different exception).
1528 This can break existing code, but in such cases the code was probably
1529 working before by accident. In the Python source, the only case of
1530 breakage discovered was in a test accidentally relying on a damaged
1531 module remaining in sys.modules. Cases are also known where tests
1532 deliberately provoking import errors remove damaged modules from
1533 sys.modules themselves, and such tests will break now if they do an
1534 unconditional del sys.modules[M].
1536 - u'%s' % obj will now try obj.__unicode__() first and fallback to
1537 obj.__str__() if no __unicode__ method can be found.
1539 - Patch #550732: Add PyArg_VaParseTupleAndKeywords(). Analogous to
1540 PyArg_VaParse(). Both are now documented. Thanks Greg Chapman.
1542 - Allow string and unicode return types from .encode()/.decode()
1543 methods on string and unicode objects. Added unicode.decode()
1544 which was missing for no apparent reason.
1546 - An attempt to fix the mess that is Python's behaviour with
1547 signal handlers and threads, complicated by readline's behaviour.
1548 It's quite possible that there are still bugs here.
1550 - Added C macros Py_CLEAR and Py_VISIT to ease the implementation of
1551 types that support garbage collection.
1553 - Compiler now treats None as a constant.
1555 - The type of values returned by __int__, __float__, __long__,
1556 __oct__, and __hex__ are now checked. Returning an invalid type
1557 will cause a TypeError to be raised. This matches the behavior of
1560 - Implemented bind_textdomain_codeset() in locale module.
1562 - Added a workaround for proper string operations in BSDs. str.split
1563 and str.is* methods can now work correctly with UTF-8 locales.
1565 - Bug #989185: unicode.iswide() and unicode.width() is dropped and
1566 the East Asian Width support is moved to unicodedata extension
1569 - Patch #941229: The source code encoding in interactive mode
1570 now refers sys.stdin.encoding not just ISO-8859-1 anymore. This
1571 allows for non-latin-1 users to write unicode strings directly.
1576 - cpickle now supports the same keyword arguments as pickle.
1581 - Added new codecs and aliases for ISO_8859-11, ISO_8859-16 and
1584 - Thanks to Edward Loper, doctest has been massively refactored, and
1585 many new features were added. Full docs will appear later. For now
1586 the doctest module comments and new test cases give good coverage.
1587 The refactoring provides many hook points for customizing behavior
1588 (such as how to report errors, and how to compare expected to actual
1589 output). New features include a <BLANKLINE> marker for expected
1590 output containing blank lines, options to produce unified or context
1591 diffs when actual output doesn't match expectations, an option to
1592 normalize whitespace before comparing, and an option to use an
1593 ellipsis to signify "don't care" regions of output.
1595 - Tkinter now supports the wish -sync and -use options.
1597 - The following methods in time support passing of None: ctime(), gmtime(),
1598 and localtime(). If None is provided, the current time is used (the
1599 same as when the argument is omitted).
1600 [SF bug 658254, patch 663482]
1602 - nntplib does now allow to ignore a .netrc file.
1604 - urllib2 now recognizes Basic authentication even if other authentication
1605 schemes are offered.
1607 - Bug #1001053. wave.open() now accepts unicode filenames.
1609 - gzip.GzipFile has a new fileno() method, to retrieve the handle of the
1610 underlying file object (provided it has a fileno() method). This is
1611 needed if you want to use os.fsync() on a GzipFile.
1613 - imaplib has two new methods: deleteacl and myrights.
1615 - nntplib has two new methods: description and descriptions. They
1616 use a more RFC-compliant way of getting a newsgroup description.
1618 - Bug #993394. Fix a possible red herring of KeyError in 'threading' being
1619 raised during interpreter shutdown from a registered function with atexit
1620 when dummy_threading is being used.
1622 - Bug #857297/Patch #916874. Fix an error when extracting a hard link
1625 - Patch #846659. Fix an error in tarfile.py when using
1626 GNU longname/longlink creation.
1628 - The obsolete FCNTL.py has been deleted. The builtin fcntl module
1629 has been available (on platforms that support fcntl) since Python
1630 1.5a3, and all FCNTL.py did is export fcntl's names, after generating
1631 a deprecation warning telling you to use fcntl directly.
1633 - Several new unicode codecs are added: big5hkscs, euc_jis_2004,
1634 iso2022_jp_2004, shift_jis_2004.
1636 - Bug #788520. Queue.{get, get_nowait, put, put_nowait} have new
1637 implementations, exploiting Conditions (which didn't exist at the time
1638 Queue was introduced). A minor semantic change is that the Full and
1639 Empty exceptions raised by non-blocking calls now occur only if the
1640 queue truly was full or empty at the instant the queue was checked (of
1641 course the Queue may no longer be full or empty by the time a calling
1642 thread sees those exceptions, though). Before, the exceptions could
1643 also be raised if it was "merely inconvenient" for the implementation
1644 to determine the true state of the Queue (because the Queue was locked
1645 by some other method in progress).
1647 - Bugs #979794 and #980117: difflib.get_grouped_opcodes() now handles the
1648 case of comparing two empty lists. This affected both context_diff() and
1651 - Bug #980938: smtplib now prints debug output to sys.stderr.
1653 - Bug #930024: posixpath.realpath() now handles infinite loops in symlinks by
1654 returning the last point in the path that was not part of any loop. Thanks
1657 - Bug #980327: ntpath not handles compressing erroneous slashes between the
1658 drive letter and the rest of the path. Also clearly handles UNC addresses now
1659 as well. Thanks Paul Moore.
1661 - bug #679953: zipfile.py should now work for files over 2 GB. The packed data
1662 for file sizes (compressed and uncompressed) was being stored as signed
1663 instead of unsigned.
1665 - decimal.py now only uses signals in the IBM spec. The other conditions are
1666 no longer part of the public API.
1668 - codecs module now has two new generic APIs: encode() and decode()
1669 which don't restrict the return types (unlike the unicode and
1670 string methods of the same name).
1672 - Non-blocking SSL sockets work again; they were broken in Python 2.3.
1675 - doctest unittest integration improvements:
1677 o Improved the unitest test output for doctest-based unit tests
1679 o Can now pass setUp and tearDown functions when creating
1682 - The threading module has a new class, local, for creating objects
1683 that provide thread-local data.
1685 - Bug #990307: when keep_empty_values is True, cgi.parse_qsl()
1686 no longer returns spurious empty fields.
1688 - Implemented bind_textdomain_codeset() in gettext module.
1690 - Introduced in gettext module the l*gettext() family of functions,
1691 which return translation strings encoded in the preferred encoding,
1692 as informed by locale module's getpreferredencoding().
1694 - optparse module (and tests) upgraded to Optik 1.5a1. Changes:
1696 - Add expansion of default values in help text: the string
1697 "%default" in an option's help string is expanded to str() of
1698 that option's default value, or "none" if no default value.
1700 - Bug #955889: option default values that happen to be strings are
1701 now processed in the same way as values from the command line; this
1702 allows generation of nicer help when using custom types. Can
1703 be disabled with parser.set_process_default_values(False).
1705 - Bug #960515: don't crash when generating help for callback
1706 options that specify 'type', but not 'dest' or 'metavar'.
1708 - Feature #815264: change the default help format for short options
1709 that take an argument from e.g. "-oARG" to "-o ARG"; add
1710 set_short_opt_delimiter() and set_long_opt_delimiter() methods to
1711 HelpFormatter to allow (slight) customization of the formatting.
1713 - Patch #736940: internationalize Optik: all built-in user-
1714 targeted literal strings are passed through gettext.gettext(). (If
1715 you want translations (.po files), they're not included with Python
1716 -- you'll find them in the Optik source distribution from
1717 http://optik.sourceforge.net/ .)
1719 - Bug #878453: respect $COLUMNS environment variable for
1720 wrapping help output.
1722 - Feature #988122: expand "%prog" in the 'description' passed
1723 to OptionParser, just like in the 'usage' and 'version' strings.
1724 (This is *not* done in the 'description' passed to OptionGroup.)
1729 - PyImport_ExecCodeModule() and PyImport_ExecCodeModuleEx(): if an
1730 error occurs while loading the module, these now delete the module's
1731 entry from sys.modules. All ways of loading modules eventually call
1732 one of these, so this is an error-case change in semantics for all
1733 ways of loading modules. In rare cases, a module loader may wish
1734 to keep a module object in sys.modules despite that the module's
1735 code cannot be executed. In such cases, the module loader must
1736 arrange to reinsert the name and module object in sys.modules.
1737 PyImport_ReloadModule() has been changed to reinsert the original
1738 module object into sys.modules if the module reload fails, so that
1739 its visible semantics have not changed.
1741 - A large pile of datetime field-extraction macros is now documented,
1742 thanks to Anthony Tuininga (patch #986010).
1747 - Improved the tutorial on creating types in C.
1749 - point out the importance of reassigning data members before
1750 assigning their values
1752 - correct my misconception about return values from visitprocs. Sigh.
1754 - mention the labor saving Py_VISIT and Py_CLEAR macros.
1756 - Major rewrite of the math module docs, to address common confusions.
1761 - The test data files for the decimal test suite are now installed on
1762 platforms that use the Makefile.
1764 - SF patch 995225: The test file testtar.tar accidentally contained
1765 CVS keywords (like $Id$), which could cause spurious failures in
1766 test_tarfile.py depending on how the test file was checked out.
1769 What's New in Python 2.4 alpha 1?
1770 =================================
1772 *Release date: 08-JUL-2004*
1777 - weakref.ref is now the type object also known as
1778 weakref.ReferenceType; it can be subclassed like any other new-style
1779 class. There's less per-entry overhead in WeakValueDictionary
1780 objects now (one object instead of three).
1782 - Bug #951851: Python crashed when reading import table of certain
1785 - Bug #215126. The locals argument to eval(), execfile(), and exec now
1786 accept any mapping type.
1788 - marshal now shares interned strings. This change introduces
1791 - Bug #966623. classes created with type() in an exec(, {}) don't
1792 have a __module__, but code in typeobject assumed it would always
1795 - Python no longer relies on the LC_NUMERIC locale setting to be
1796 the "C" locale; as a result, it no longer tries to prevent changing
1797 the LC_NUMERIC category.
1799 - Bug #952807: Unpickling pickled instances of subclasses of
1800 datetime.date, datetime.datetime and datetime.time could yield insane
1801 objects. Thanks to Jiwon Seo for a fix.
1803 - Bug #845802: Python crashes when __init__.py is a directory.
1805 - Unicode objects received two new methods: iswide() and width().
1806 These query East Asian width information, as specified in Unicode
1809 - Improved the tuple hashing algorithm to give fewer collisions in
1810 common cases. Fixes bug #942952.
1812 - Implemented generator expressions (PEP 289). Coded by Jiwon Seo.
1814 - Enabled the profiling of C extension functions (and builtins) - check
1815 new documentation and modified profile and bdb modules for more details
1817 - Set file.name to the object passed to open (instead of a new string)
1819 - Moved tracebackobject into traceback.h and renamed to PyTracebackObject
1821 - Optimized the byte coding for multiple assignments like "a,b=b,a" and
1822 "a,b,c=1,2,3". Improves their speed by 25% to 30%.
1824 - Limit the nested depth of a tuple for the second argument to isinstance()
1825 and issubclass() to the recursion limit of the interpreter.
1828 - Optimized dict iterators, creating separate types for each
1829 and having them reveal their length. Also optimized the
1830 methods: keys(), values(), and items().
1832 - Implemented a newcode opcode, LIST_APPEND, that simplifies
1833 the generated bytecode for list comprehensions and further
1834 improves their performance (about 35%).
1836 - Implemented rich comparisons for floats, which seems to make
1837 comparisons involving NaNs somewhat less surprising when the
1838 underlying C compiler actually implements C99 semantics.
1840 - Optimized list.extend() to save memory and no longer create
1841 intermediate sequences. Also, extend() now pre-allocates the
1842 needed memory whenever the length of the iterable is known in
1843 advance -- this halves the time to extend the list.
1845 - Optimized list resize operations to make fewer calls to the system
1846 realloc(). Significantly speeds up list appends, list pops,
1847 list comprehensions, and the list constructor (when the input iterable
1848 length is not known).
1850 - Changed the internal list over-allocation scheme. For larger lists,
1851 overallocation ranged between 3% and 25%. Now, it is a constant 12%.
1852 For smaller lists (n<8), overallocation was upto eight elements. Now,
1853 the overallocation is no more than three elements -- this improves space
1854 utilization for applications that have large numbers of small lists.
1856 - Most list bodies now get re-used rather than freed. Speeds up list
1857 instantiation and deletion by saving calls to malloc() and free().
1859 - The dict.update() method now accepts all the same argument forms
1860 as the dict() constructor. This now includes item lists and/or
1863 - Support for arbitrary objects supporting the read-only buffer
1864 interface as the co_code field of code objects (something that was
1865 only possible to create from C code) has been removed.
1867 - Made omitted callback and None equivalent for weakref.ref() and
1868 weakref.proxy(); the None case wasn't handled correctly in all
1871 - Fixed problem where PyWeakref_NewRef() and PyWeakref_NewProxy()
1872 assumed that initial existing entries in an object's weakref list
1873 would not be removed while allocating a new weakref object. Since
1874 GC could be invoked at that time, however, that assumption was
1875 invalid. In a truly obscure case of GC being triggered during
1876 creation for a new weakref object for an referent which already
1877 has a weakref without a callback which is only referenced from
1878 cyclic trash, a memory error can occur. This consistently created a
1879 segfault in a debug build, but provided less predictable behavior in
1882 - input() builtin function now respects compiler flags such as
1883 __future__ statements. SF patch 876178.
1885 - Removed PendingDeprecationWarning from apply(). apply() remains
1886 deprecated, but the nuisance warning will not be issued.
1888 - At Python shutdown time (Py_Finalize()), 2.3 called cyclic garbage
1889 collection twice, both before and after tearing down modules. The
1890 call after tearing down modules has been disabled, because too much
1891 of Python has been torn down then for __del__ methods and weakref
1892 callbacks to execute sanely. The most common symptom was a sequence
1893 of uninformative messages on stderr when Python shut down, produced
1894 by threads trying to raise exceptions, but unable to report the nature
1895 of their problems because too much of the sys module had already been
1898 - Removed FutureWarnings related to hex/oct literals and conversions
1899 and left shifts. (Thanks to Kalle Svensson for SF patch 849227.)
1900 This addresses most of the remaining semantic changes promised by
1901 PEP 237, except for repr() of a long, which still shows the trailing
1902 'L'. The PEP appears to promise warnings for operations that
1903 changed semantics compared to Python 2.3, but this is not
1904 implemented; we've suffered through enough warnings related to
1905 hex/oct literals and I think it's best to be silent now.
1907 - For str and unicode objects, the ljust(), center(), and rjust()
1908 methods now accept an optional argument specifying a fill
1909 character other than a space.
1911 - When method objects have an attribute that can be satisfied either
1912 by the function object or by the method object, the function
1913 object's attribute usually wins. Christian Tismer pointed out that
1914 that this is really a mistake, because this only happens for special
1915 methods (like __reduce__) where the method object's version is
1916 really more appropriate than the function's attribute. So from now
1917 on, all method attributes will have precedence over function
1918 attributes with the same name.
1920 - Critical bugfix, for SF bug 839548: if a weakref with a callback,
1921 its callback, and its weakly referenced object, all became part of
1922 cyclic garbage during a single run of garbage collection, the order
1923 in which they were torn down was unpredictable. It was possible for
1924 the callback to see partially-torn-down objects, leading to immediate
1925 segfaults, or, if the callback resurrected garbage objects, to
1926 resurrect insane objects that caused segfaults (or other surprises)
1927 later. In one sense this wasn't surprising, because Python's cyclic gc
1928 had no knowledge of Python's weakref objects. It does now. When
1929 weakrefs with callbacks become part of cyclic garbage now, those
1930 weakrefs are cleared first. The callbacks don't trigger then,
1931 preventing the problems. If you need callbacks to trigger, then just
1932 as when cyclic gc is not involved, you need to write your code so
1933 that weakref objects outlive the objects they weakly reference.
1935 - Critical bugfix, for SF bug 840829: if cyclic garbage collection
1936 happened to occur during a weakref callback for a new-style class
1937 instance, subtle memory corruption was the result (in a release build;
1938 in a debug build, a segfault occurred reliably very soon after).
1939 This has been repaired.
1941 - Compiler flags set in PYTHONSTARTUP are now active in __main__.
1943 - Added two builtin types, set() and frozenset().
1945 - Added a reversed() builtin function that returns a reverse iterator
1948 - Added a sorted() builtin function that returns a new sorted list
1951 - CObjects are now mutable (on the C level) through PyCObject_SetVoidPtr.
1953 - list.sort() now supports three keyword arguments: cmp, key, and reverse.
1954 The key argument can be a function of one argument that extracts a
1955 comparison key from the original record: mylist.sort(key=str.lower).
1956 The reverse argument is a boolean value and if True will change the
1957 sort order as if the comparison arguments were reversed. In addition,
1958 the documentation has been amended to provide a guarantee that all sorts
1959 starting with Py2.3 are guaranteed to be stable (the relative order of
1960 records with equal keys is unchanged).
1962 - Added test whether wchar_t is signed or not. A signed wchar_t is not
1963 usable as internal unicode type base for Py_UNICODE since the
1964 unicode implementation assumes an unsigned type.
1966 - Fixed a bug in the cache of length-one Unicode strings that could
1967 lead to a seg fault. The specific problem occurred when an earlier,
1968 non-fatal error left an uninitialized Unicode object in the
1971 - The % formatting operator now supports '%F' which is equivalent to
1972 '%f'. This has always been documented but never implemented.
1974 - complex(obj) could leak a little memory if obj wasn't a string or
1977 - zip() with no arguments now returns an empty list instead of raising
1978 a TypeError exception.
1980 - obj.__contains__() now returns True/False instead of 1/0. SF patch
1983 - Python no longer tries to be smart about recursive comparisons.
1984 When comparing containers with cyclic references to themselves it
1985 will now just hit the recursion limit. See SF patch 825639.
1987 - str and unicode builtin types now have an rsplit() method that is
1988 same as split() except that it scans the string from the end
1989 working towards the beginning. See SF feature request 801847.
1991 - Fixed a bug in object.__reduce_ex__ when using protocol 2. Failure
1992 to clear the error when attempts to get the __getstate__ attribute
1993 fail caused intermittent errors and odd behavior.
1995 - buffer objects based on other objects no longer cache a pointer to
1996 the data and the data length. Instead, the appropriate tp_as_buffer
1997 method is called as necessary.
1999 - fixed: if a file is opened with an explicit buffer size >= 1, repeated
2000 close() calls would attempt to free() the buffer already free()ed on
2007 - Added socket.getservbyport(), and make the second argument in
2008 getservbyname() and getservbyport() optional.
2010 - time module code that deals with input POSIX timestamps will now raise
2011 ValueError if more than a second is lost in precision when the
2012 timestamp is cast to the platform C time_t type. There's no chance
2013 that the platform will do anything sensible with the result in such
2014 cases. This includes ctime(), localtime() and gmtime(). Assorted
2015 fromtimestamp() and utcfromtimestamp() methods in the datetime module
2016 were also protected. Closes bugs #919012 and 975996.
2018 - fcntl.ioctl now warns if the mutate flag is not specified.
2020 - nt now properly allows to refer to UNC roots, e.g. in nt.stat().
2022 - the weakref module now supports additional objects: array.array,
2023 sre.pattern_objects, file objects, and sockets.
2025 - operator.isMappingType() and operator.isSequenceType() now give
2026 fewer false positives.
2028 - socket.sslerror is now a subclass of socket.error . Also added
2029 socket.error to the socket module's C API.
2031 - Bug #920575: A problem where the _locale module segfaults on
2032 nl_langinfo(ERA) caused by GNU libc's illegal NULL return is fixed.
2034 - array objects now support the copy module. Also, their resizing
2035 scheme has been updated to match that used for list objects. This improves
2036 the performance (speed and memory usage) of append() operations.
2037 Also, array.array() and array.extend() now accept any iterable argument
2038 for repeated appends without needing to create another temporary array.
2040 - cStringIO.writelines() now accepts any iterable argument and writes
2041 the lines one at a time rather than joining them and writing once.
2042 Made a parallel change to StringIO.writelines(). Saves memory and
2043 makes suitable for use with generator expressions.
2045 - time.strftime() now checks that the values in its time tuple argument
2046 are within the proper boundaries to prevent possible crashes from the
2047 platform's C library implementation of strftime(). Can possibly
2048 break code that uses values outside the range that didn't cause
2049 problems previously (such as sitting day of year to 0). Fixes bug
2052 - The socket module now supports Bluetooth sockets, if the
2053 system has <bluetooth/bluetooth.h>
2055 - Added a collections module containing a new datatype, deque(),
2056 offering high-performance, thread-safe, memory friendly appends
2057 and pops on either side of the deque.
2059 - Several modules now take advantage of collections.deque() for
2060 improved performance: Queue, mutex, shlex, threading, and pydoc.
2062 - The operator module has two new functions, attrgetter() and
2063 itemgetter() which are useful for creating fast data extractor
2064 functions for map(), list.sort(), itertools.groupby(), and
2065 other functions that expect a function argument.
2067 - socket.SHUT_{RD,WR,RDWR} was added.
2069 - os.getsid was added.
2071 - The pwd module incorrectly advertised its struct type as
2072 struct_pwent; this has been renamed to struct_passwd. (The old name
2073 is still supported for backwards compatibility.)
2075 - The xml.parsers.expat module now provides Expat 1.95.7.
2077 - socket.IPPROTO_IPV6 was added.
2079 - readline.clear_history was added.
2081 - select.select() now accepts sequences for its first three arguments.
2083 - cStringIO now supports the f.closed attribute.
2085 - The signal module now exposes SIGRTMIN and SIGRTMAX (if available).
2087 - curses module now supports use_default_colors(). [patch #739124]
2089 - Bug #811028: ncurses.h breakage on FreeBSD/MacOS X
2091 - Bug #814613: INET_ADDRSTRLEN fix needed for all compilers on SGI
2093 - Implemented non-recursive SRE matching scheme (#757624).
2095 - Implemented (?(id/name)yes|no) support in SRE (#572936).
2097 - random.seed() with no arguments or None uses time.time() as a default
2098 seed. Modified to match Py2.2 behavior and use fractional seconds so
2099 that successive runs are more likely to produce different sequences.
2101 - random.Random has a new method, getrandbits(k), which returns an int
2102 with k random bits. This method is now an optional part of the API
2103 for user defined generators. Any generator that defines genrandbits()
2104 can now use randrange() for ranges with a length >= 2**53. Formerly,
2105 randrange would return only even numbers for ranges that large (see
2106 SF bug #812202). Generators that do not define genrandbits() now
2107 issue a warning when randrange() is called with a range that large.
2109 - itertools has a new function, groupby() for aggregating iterables
2110 into groups sharing the same key (as determined by a key function).
2111 It offers some of functionality of SQL's groupby keyword and of
2112 the Unix uniq filter.
2114 - itertools now has a new tee() function which produces two independent
2115 iterators from a single iterable.
2117 - itertools.izip() with no arguments now returns an empty iterator instead
2118 of raising a TypeError exception.
2120 - Fixed #853061: allow BZ2Compressor.compress() to receive an empty string
2126 - Added a new module: cProfile, a C profiler with the same interface as the
2127 profile module. cProfile avoids some of the drawbacks of the hotshot
2128 profiler and provides a bit more information than the other two profilers.
2129 Based on "lsprof" (patch #1212837).
2131 - Bug #1266283: The new function "lexists" is now in os.path.__all__.
2133 - Bug #981530: Fix UnboundLocalError in shutil.rmtree(). This affects
2134 the documented behavior: the function passed to the onerror()
2135 handler can now also be os.listdir.
2137 - Bug #754449: threading.Thread objects no longer mask exceptions raised during
2138 interpreter shutdown with another exception from attempting to handle the
2141 - Added decimal.py per PEP 327.
2143 - Bug #981299: rsync is now a recognized protocol in urlparse that uses a
2144 "netloc" portion of a URL.
2146 - Bug #919012: shutil.move() will not try to move a directory into itself.
2147 Thanks Johannes Gijsbers.
2149 - Bug #934282: pydoc.stripid() is now case-insensitive. Thanks Robin Becker.
2151 - Bug #823209: cmath.log() now takes an optional base argument so that its
2152 API matches math.log().
2154 - Bug #957381: distutils bdist_rpm no longer fails on recent RPM versions
2155 that generate a -debuginfo.rpm
2157 - os.path.devnull has been added for all supported platforms.
2159 - Fixed #877165: distutils now picks the right C++ compiler command
2160 on cygwin and mingw32.
2162 - urllib.urlopen().readline() now handles HTTP/0.9 correctly.
2164 - refactored site.py into functions. Also wrote regression tests for the
2167 - The distutils install command now supports the --home option and
2168 installation scheme for all platforms.
2170 - asyncore.loop now has a repeat count parameter that defaults to
2173 - The distutils sdist command now ignores all .svn directories, in
2174 addition to CVS and RCS directories. .svn directories hold
2175 administrative files for the Subversion source control system.
2177 - Added a new module: cookielib. Automatic cookie handling for HTTP
2178 clients. Also, support for cookielib has been added to urllib2, so
2179 urllib2.urlopen() can transparently handle cookies.
2181 - stringprep.py now uses built-in set() instead of sets.Set().
2183 - Bug #876278: Unbounded recursion in modulefinder
2185 - Bug #780300: Swap public and system ID in LexicalHandler.startDTD.
2186 Applications relying on the wrong order need to be corrected.
2188 - Bug #926075: Fixed a bug that returns a wrong pattern object
2189 for a string or unicode object in sre.compile() when a different
2190 type pattern with the same value exists.
2192 - Added countcallers arg to trace.Trace class (--trackcalls command line arg
2193 when run from the command prompt).
2195 - Fixed a caching bug in platform.platform() where the argument of 'terse' was
2196 not taken into consideration when caching value.
2198 - Added two new command-line arguments for profile (output file and
2201 - Added global runctx function to profile module
2203 - Add hlist missing entryconfigure and entrycget methods.
2205 - The ptcp154 codec was added for Kazakh character set support.
2207 - Support non-anonymous ftp URLs in urllib2.
2209 - The encodings package will now apply codec name aliases
2210 first before starting to try the import of the codec module.
2211 This simplifies overriding built-in codecs with external
2212 packages, e.g. the included CJK codecs with the JapaneseCodecs
2213 package, by adjusting the aliases dictionary in encodings.aliases
2216 - base64 now supports RFC 3548 Base16, Base32, and Base64 encoding and
2219 - urllib2 now supports processors. A processor is a handler that
2220 implements an xxx_request or xxx_response method. These methods are
2221 called for all requests.
2223 - distutils compilers now compile source files in the same order as
2224 they are passed to the compiler.
2226 - pprint.pprint() and pprint.pformat() now have additional parameters
2227 indent, width and depth.
2229 - Patch #750542: pprint now will pretty print subclasses of list, tuple
2230 and dict too, as long as they don't overwrite __repr__().
2232 - Bug #848614: distutils' msvccompiler fails to find the MSVC6
2233 compiler because of incomplete registry entries.
2235 - httplib.HTTP.putrequest now offers to omit the implicit Accept-Encoding.
2237 - Patch #841977: modulefinder didn't find extension modules in packages
2239 - imaplib.IMAP4.thread was added.
2241 - Plugged a minor hole in tempfile.mktemp() due to the use of
2242 os.path.exists(), switched to using os.lstat() directly if possible.
2244 - bisect.py and heapq.py now have underlying C implementations
2245 for better performance.
2247 - heapq.py has two new functions, nsmallest() and nlargest().
2249 - traceback.format_exc has been added (similar to print_exc but it returns
2252 - xmlrpclib.MultiCall has been added.
2254 - poplib.POP3_SSL has been added.
2256 - tmpfile.mkstemp now returns an absolute path even if dir is relative.
2258 - urlparse is RFC 2396 compliant.
2260 - The fieldnames argument to the csv module's DictReader constructor is now
2261 optional. If omitted, the first row of the file will be used as the
2264 - encodings.bz2_codec was added for access to bz2 compression
2265 using "a long string".encode('bz2')
2267 - Various improvements to unittest.py, realigned with PyUnit CVS.
2269 - dircache now passes exceptions to the caller, instead of returning
2272 - The bsddb module and dbhash module now support the iterator and
2273 mapping protocols which make them more substitutable for dictionaries
2276 - The csv module's DictReader and DictWriter classes now accept keyword
2277 arguments. This was an omission in the initial implementation.
2279 - The email package handles some RFC 2231 parameters with missing
2280 CHARSET fields better. It also includes a patch to parameter
2281 parsing when semicolons appear inside quotes.
2283 - sets.py now runs under Py2.2. In addition, the argument restrictions
2284 for most set methods (but not the operators) have been relaxed to
2287 - _strptime.py now has a behind-the-scenes caching mechanism for the most
2288 recent TimeRE instance used along with the last five unique directive
2289 patterns. The overall module was also made more thread-safe.
2291 - random.cunifvariate() and random.stdgamma() were deprecated in Py2.3
2292 and removed in Py2.4.
2294 - Bug #823328: urllib2.py's HTTP Digest Auth support works again.
2296 - Patch #873597: CJK codecs are imported into rank of default codecs.
2301 - A hotshotmain script was added to the Tools/scripts directory that
2302 makes it easy to run a script under control of the hotshot profiler.
2304 - The db2pickle and pickle2db scripts can now dump/load gdbm files.
2306 - The file order on the command line of the pickle2db script was reversed.
2307 It is now [ picklefile ] dbfile. This provides better symmetry with
2308 db2pickle. The file arguments to both scripts are now source followed by
2309 destination in situations where both files are given.
2311 - The pydoc script will display a link to the module documentation for
2312 modules determined to be part of the core distribution. The documentation
2313 base directory defaults to http://www.python.org/doc/current/lib/ but can
2314 be changed by setting the PYTHONDOCS environment variable.
2316 - texcheck.py now detects double word errors.
2318 - md5sum.py mistakenly opened input files in text mode by default, a
2319 silent and dangerous change from previous releases. It once again
2320 opens input files in binary mode by default. The -t and -b flags
2321 remain for compatibility with the 2.3 release, but -b is the default
2324 - py-electric-colon now works when pending-delete/delete-selection mode is
2327 - py-help-at-point is no longer bound to the F1 key - it's still bound to
2330 - Pynche was fixed to not crash when there is no ~/.pynche file and no
2331 -d option was given.
2336 - Bug #978645: Modules/getpath.c now builds properly in --disable-framework
2339 - Profiling using gprof is now available if Python is configured with
2342 - Profiling the VM using the Pentium TSC is now possible if Python
2343 is configured --with-tsc.
2345 - In order to find libraries, setup.py now also looks in /lib64, for use
2348 - Bug #934635: Fixed a bug where the configure script couldn't detect
2349 getaddrinfo() properly if the KAME stack had SCTP support.
2351 - Support for missing ANSI C header files (limits.h, stddef.h, etc) was
2354 - Systems requiring the D4, D6 or D7 variants of pthreads are no longer
2355 supported (see PEP 11).
2357 - Universal newline support can no longer be disabled (see PEP 11).
2359 - Support for DGUX, SunOS 4, IRIX 4 and Minix was removed (see PEP 11).
2361 - Support for systems requiring --with-dl-dld or --with-sgi-dl was removed
2364 - Tests for sizeof(char) were removed since ANSI C mandates that
2365 sizeof(char) must be 1.
2370 - Thanks to Anthony Tuininga, the datetime module now supplies a C API
2371 containing type-check macros and constructors. See new docs in the
2372 Python/C API Reference Manual for details.
2374 - Private function _PyTime_DoubleToTimet added, to convert a Python
2375 timestamp (C double) to platform time_t with some out-of-bounds
2376 checking. Declared in new header file timefuncs.h. It would be
2377 good to expose some other internal timemodule.c functions there.
2379 - New public functions PyEval_EvaluateFrame and PyGen_New to expose
2382 - New public functions Py_IncRef() and Py_DecRef(), exposing the
2383 functionality of the Py_XINCREF() and Py_XDECREF macros. Useful for
2384 runtime dynamic embedding of Python. See patch #938302, by Bob
2387 - Added a new macro, PySequence_Fast_ITEMS, which retrieves a fast sequence's
2388 underlying array of PyObject pointers. Useful for high speed looping.
2390 - Created a new method flag, METH_COEXIST, which causes a method to be loaded
2391 even if already defined by a slot wrapper. This allows a __contains__
2392 method, for example, to co-exist with a defined sq_contains slot. This
2393 is helpful because the PyCFunction can take advantage of optimized calls
2394 whenever METH_O or METH_NOARGS flags are defined.
2396 - Added a new function, PyDict_Contains(d, k) which is like
2397 PySequence_Contains() but is specific to dictionaries and executes
2400 - Added three new macros: Py_RETURN_NONE, Py_RETURN_TRUE, and Py_RETURN_FALSE.
2401 Each return the singleton they mention after Py_INCREF()ing them.
2403 - Added a new function, PyTuple_Pack(n, ...) for constructing tuples from a
2404 variable length argument list of Python objects without having to invoke
2405 the more complex machinery of Py_BuildValue(). PyTuple_Pack(3, a, b, c)
2406 is equivalent to Py_BuildValue("(OOO)", a, b, c).
2411 - The _winreg module could segfault when reading very large registry
2412 values, due to unchecked alloca() calls (SF bug 851056). The fix is
2413 uses either PyMem_Malloc(n) or PyString_FromStringAndSize(NULL, n),
2414 as appropriate, followed by a size check.
2416 - file.truncate() could misbehave if the file was open for update
2417 (modes r+, rb+, w+, wb+), and the most recent file operation before
2418 the truncate() call was an input operation. SF bug 801631.
2421 What's New in Python 2.3 final?
2422 ===============================
2424 *Release date: 29-Jul-2003*
2429 - Bug 778400: IDLE hangs when selecting "Edit with IDLE" from explorer.
2430 This was unique to Windows, and was fixed by adding an -n switch to
2431 the command the Windows installer creates to execute "Edit with IDLE"
2432 context-menu actions.
2434 - IDLE displays a new message upon startup: some "personal firewall"
2435 kinds of programs (for example, ZoneAlarm) open a dialog of their
2436 own when any program opens a socket. IDLE does use sockets, talking
2437 on the computer's internal loopback interface. This connection is not
2438 visible on any external interface and no data is sent to or received
2439 from the Internet. So, if you get such a dialog when opening IDLE,
2440 asking whether to let pythonw.exe talk to address 127.0.0.1, say yes,
2441 and rest assured no communication external to your machine is taking
2442 place. If you don't allow it, IDLE won't be able to start.
2445 What's New in Python 2.3 release candidate 2?
2446 =============================================
2448 *Release date: 24-Jul-2003*
2453 - It is now possible to import from zipfiles containing additional
2454 data bytes before the zip compatible archive. Zipfiles containing a
2455 comment at the end are still unsupported.
2460 - A longstanding bug in the parser module's initialization could cause
2461 fatal internal refcount confusion when the module got initialized more
2462 than once. This has been fixed.
2464 - Fixed memory leak in pyexpat; using the parser's ParseFile() method
2465 with open files that aren't instances of the standard file type
2466 caused an instance of the bound .read() method to be leaked on every
2469 - Fixed some leaks in the locale module.
2474 - Lib/encodings/rot_13.py when used as a script, now more properly
2475 uses the first Python interpreter on your path.
2477 - Removed caching of TimeRE (and thus LocaleTime) in _strptime.py to
2478 fix a locale related bug in the test suite. Although another patch
2479 was needed to actually fix the problem, the cache code was not
2490 - For MacOSX, added -mno-fused-madd to BASECFLAGS to fix test_coercion
2491 on Panther (OSX 10.3).
2499 - The tempfile module could do insane imports on Windows if PYTHONCASEOK
2500 was set, making temp file creation impossible. Repaired.
2502 - Add a patch to workaround pthread_sigmask() bugs in Cygwin.
2507 - Various fixes to pimp.
2509 - Scripts runs with pythonw no longer had full window manager access.
2511 - Don't force boot-disk-only install, for reasons unknown it causes
2512 more problems than it solves.
2515 What's New in Python 2.3 release candidate 1?
2516 =============================================
2518 *Release date: 18-Jul-2003*
2523 - The new function sys.getcheckinterval() returns the last value set
2524 by sys.setcheckinterval().
2526 - Several bugs in the symbol table phase of the compiler have been
2527 fixed. Errors could be lost and compilation could fail without
2528 reporting an error. SF patch 763201.
2530 - The interpreter is now more robust about importing the warnings
2531 module. In an executable generated by freeze or similar programs,
2532 earlier versions of 2.3 would fail if the warnings module could
2533 not be found on the file system. Fixes SF bug 771097.
2535 - A warning about assignments to module attributes that shadow
2536 builtins, present in earlier releases of 2.3, has been removed.
2538 - It is not possible to create subclasses of builtin types like str
2539 and tuple that define an itemsize. Earlier releases of Python 2.3
2540 allowed this by mistake, leading to crashes and other problems.
2542 - The thread_id is now initialized to 0 in a non-thread build. SF bug
2545 - SF bug 762891: "del p[key]" on proxy object no longer raises SystemError.
2550 - weakref.proxy() can now handle "del obj[i]" for proxy objects
2551 defining __delitem__. Formerly, it generated a SystemError.
2553 - SSL no longer crashes the interpreter when the remote side disconnects.
2555 - On Unix the mmap module can again be used to map device files.
2557 - time.strptime now exclusively uses the Python implementation
2558 contained within the _strptime module.
2560 - The print slot of weakref proxy objects was removed, because it was
2561 not consistent with the object's repr slot.
2563 - The mmap module only checks file size for regular files, not
2564 character or block devices. SF patch 708374.
2566 - The cPickle Pickler garbage collection support was fixed to traverse
2567 the find_class attribute, if present.
2569 - There are several fixes for the bsddb3 wrapper module.
2571 bsddb3 no longer crashes if an environment is closed before a cursor
2574 The DB and DBEnv set_get_returns_none function was extended to take
2575 a level instead of a boolean flag. The new level 2 means that in
2576 addition, cursor.set()/.get() methods return None instead of raising
2579 A typo was fixed in DBCursor.join_item(), preventing a crash.
2584 - distutils now supports MSVC 7.1
2586 - doctest now examines all docstrings by default. Previously, it would
2587 skip over functions with private names (as indicated by the underscore
2588 naming convention). The old default created too much of a risk that
2589 user tests were being skipped inadvertently. Note, this change could
2590 break code in the unlikely case that someone had intentionally put
2591 failing tests in the docstrings of private functions. The breakage
2592 is easily fixable by specifying the old behavior when calling testmod()
2595 - There were several fixes to the way dumbdbms are closed. It's vital
2596 that a dumbdbm database be closed properly, else the on-disk data
2597 and directory files can be left in mutually inconsistent states.
2598 dumbdbm.py's _Database.__del__() method attempted to close the
2599 database properly, but a shutdown race in _Database._commit() could
2600 prevent this from working, so that a program trusting __del__() to
2601 get the on-disk files in synch could be badly surprised. The race
2602 has been repaired. A sync() method was also added so that shelve
2603 can guarantee data is written to disk.
2605 The close() method can now be called more than once without complaint.
2607 - The classes in threading.py are now new-style classes. That they
2608 weren't before was an oversight.
2610 - The urllib2 digest authentication handlers now define the correct
2611 auth_header. The earlier versions would fail at runtime.
2613 - SF bug 763023: fix uncaught ZeroDivisionError in difflib ratio methods
2614 when there are no lines.
2616 - SF bug 763637: fix exception in Tkinter with after_cancel
2617 which could occur with Tk 8.4
2619 - SF bug 770601: CGIHTTPServer.py now passes the entire environment
2622 - SF bug 765238: add filter to fnmatch's __all__.
2624 - SF bug 748201: make time.strptime() error messages more helpful.
2626 - SF patch 764470: Do not dump the args attribute of a Fault object in
2629 - SF patch 549151: urllib and urllib2 now redirect POSTs on 301
2632 - SF patch 766650: The whichdb module was fixed to recognize dbm files
2633 generated by gdbm on OS/2 EMX.
2635 - SF bugs 763047 and 763052: fixes bug of timezone value being left as
2636 -1 when ``time.tzname[0] == time.tzname[1] and not time.daylight``
2637 is true when it should only when time.daylight is true.
2639 - SF bug 764548: re now allows subclasses of str and unicode to be
2642 - SF bug 763637: In Tkinter, change after_cancel() to handle tuples
2643 of varying sizes. Tk 8.4 returns a different number of values
2646 - SF bug 763023: difflib.ratio() did not catch zero division.
2648 - The Queue module now has an __all__ attribute.
2653 - See Lib/idlelib/NEWS.txt for IDLE news.
2655 - SF bug 753592: webchecker/wsgui now handles user supplied directories.
2657 - The trace.py script has been removed. It is now in the standard library.
2662 - Python now compiles with -fno-strict-aliasing if possible (SF bug 766696).
2664 - The socket module compiles on IRIX 6.5.10.
2666 - An irix64 system is treated the same way as an irix6 system (SF
2669 - Several definitions were missing on FreeBSD 5.x unless the
2670 __BSD_VISIBLE symbol was defined. configure now defines it as
2676 - Unicode objects now support mbcs as a built-in encoding, so the C
2677 API can use it without deferring to the encodings package.
2682 - The Windows implementation of PyThread_start_new_thread() never
2683 checked error returns from Windows functions correctly. As a result,
2684 it could claim to start a new thread even when the Microsoft
2685 _beginthread() function failed (due to "too many threads" -- this is
2686 on the order of thousands when it happens). In these cases, the
2689 thread.error: can't start new thread
2693 - SF bug 766669: Prevent a GPF on interpreter exit when sockets are in
2694 use. The interpreter now calls WSACleanup() from Py_Finalize()
2695 instead of from DLL teardown.
2700 - Bundlebuilder now inherits default values in the right way. It was
2701 previously possible for app bundles to get a type of "BNDL" instead
2702 of "APPL." Other improvements include, a --build-id option to
2703 specify the CFBundleIdentifier and using the --python option to set
2704 the executable in the bundle.
2706 - Fixed two bugs in MacOSX framework handling.
2708 - pythonw did not allow user interaction in 2.3rc1, this has been fixed.
2710 - Python is now compiled with -mno-fused-madd, making all tests pass
2713 What's New in Python 2.3 beta 2?
2714 ================================
2716 *Release date: 29-Jun-2003*
2721 - A program can now set the environment variable PYTHONINSPECT to some
2722 string value in Python, and cause the interpreter to enter the
2723 interactive prompt at program exit, as if Python had been invoked
2726 - list.index() now accepts optional start and stop arguments. Similar
2727 changes were made to UserList.index(). SF feature request 754014.
2729 - SF patch 751998 fixes an unwanted side effect of the previous fix
2730 for SF bug 742860 (the next item).
2732 - SF bug 742860: "WeakKeyDictionary __delitem__ uses iterkeys". This
2733 wasn't threadsafe, was very inefficient (expected time O(len(dict))
2734 instead of O(1)), and could raise a spurious RuntimeError if another
2735 thread mutated the dict during __delitem__, or if a comparison function
2736 mutated it. It also neglected to raise KeyError when the key wasn't
2737 present; didn't raise TypeError when the key wasn't of a weakly
2738 referencable type; and broke various more-or-less obscure dict
2739 invariants by using a sequence of equality comparisons over the whole
2740 set of dict keys instead of computing the key's hash code to narrow
2741 the search to those keys with the same hash code. All of these are
2742 considered to be bugs. A new implementation of __delitem__ repairs all
2743 that, but note that fixing these bugs may change visible behavior in
2744 code relying (whether intentionally or accidentally) on old behavior.
2746 - SF bug 734869: Fixed a compiler bug that caused a fatal error when
2747 compiling a list comprehension that contained another list comprehension
2748 embedded in a lambda expression.
2750 - SF bug 705231: builtin pow() no longer lets the platform C pow()
2751 raise -1.0 to integer powers, because (at least) glibc gets it wrong
2752 in some cases. The result should be -1.0 if the power is odd and 1.0
2753 if the power is even, and any float with a sufficiently large exponent
2754 is (mathematically) an exact even integer.
2756 - SF bug 759227: A new-style class that implements __nonzero__() must
2757 return a bool or int (but not an int subclass) from that method. This
2758 matches the restriction on classic classes.
2760 - The encoding attribute has been added for file objects, and set to
2761 the terminal encoding on Unix and Windows.
2763 - The softspace attribute of file objects became read-only by oversight.
2764 It's writable again.
2766 - Reverted a 2.3 beta 1 change to iterators for subclasses of list and
2767 tuple. By default, the iterators now access data elements directly
2768 instead of going through __getitem__. If __getitem__ access is
2769 preferred, then __iter__ can be overridden.
2771 - SF bug 735247: The staticmethod and super types participate in
2772 garbage collection. Before this change, it was possible for leaks to
2773 occur in functions with non-global free variables that used these types.
2778 - the socket module has a new exception, socket.timeout, to allow
2779 timeouts to be handled separately from other socket errors.
2781 - SF bug 751276: cPickle has fixed to propagate exceptions raised in
2782 user code. In earlier versions, cPickle caught and ignored any
2783 exception when it performed operations that it expected to raise
2784 specific exceptions like AttributeError.
2786 - cPickle Pickler and Unpickler objects now participate in garbage
2789 - mimetools.choose_boundary() could return duplicate strings at times,
2790 especially likely on Windows. The strings returned are now guaranteed
2791 unique within a single program run.
2793 - thread.interrupt_main() raises KeyboardInterrupt in the main thread.
2794 dummy_thread has also been modified to try to simulate the behavior.
2796 - array.array.insert() now treats negative indices as being relative
2797 to the end of the array, just like list.insert() does. (SF bug #739313)
2799 - The datetime module classes datetime, time, and timedelta are now
2800 properly subclassable.
2802 - _tkinter.{get|set}busywaitinterval was added.
2804 - itertools.islice() now accepts stop=None as documented.
2805 Fixes SF bug #730685.
2807 - the bsddb185 module is built in one restricted instance -
2808 /usr/include/db.h exists and defines HASHVERSION to be 2. This is true
2809 for many BSD-derived systems.
2815 - Some happy doctest extensions from Jim Fulton have been added to
2816 doctest.py. These are already being used in Zope3. The two
2819 doctest.debug(module, name) extracts the doctests from the named object
2820 in the given module, puts them in a temp file, and starts pdb running
2821 on that file. This is great when a doctest fails.
2823 doctest.DocTestSuite(module=None) returns a synthesized unittest
2824 TestSuite instance, to be run by the unittest framework, which
2825 runs all the doctests in the module. This allows writing tests in
2826 doctest style (which can be clearer and shorter than writing tests
2827 in unittest style), without losing unittest's powerful testing
2828 framework features (which doctest lacks).
2830 - For compatibility with doctests created before 2.3, if an expected
2831 output block consists solely of "1" and the actual output block
2832 consists solely of "True", it's accepted as a match; similarly
2833 for "0" and "False". This is quite un-doctest-like, but is practical.
2834 The behavior can be disabled by passing the new doctest module
2835 constant DONT_ACCEPT_TRUE_FOR_1 to the new optionflags optional
2838 - ZipFile.testzip() now only traps BadZipfile exceptions. Previously,
2839 a bare except caught to much and reported all errors as a problem
2842 - The logging module now has a new function, makeLogRecord() making
2843 LogHandler easier to interact with DatagramHandler and SocketHandler.
2845 - The cgitb module has been extended to support plain text display (SF patch
2848 - A brand new version of IDLE (from the IDLEfork project at
2849 SourceForge) is now included as Lib/idlelib. The old Tools/idle is
2852 - Added a new module: trace (documentation missing). This module used
2853 to be distributed in Tools/scripts. It uses sys.settrace() to trace
2854 code execution -- either function calls or individual lines. It can
2855 generate tracing output during execution or a post-mortem report of
2858 - The threading module has new functions settrace() and setprofile()
2859 that cooperate with the functions of the same name in the sys
2860 module. A function registered with the threading module will
2861 be used for all threads it creates. The new trace module uses this
2862 to provide tracing for code running in threads.
2864 - copy.py: applied SF patch 707900, fixing bug 702858, by Steven
2865 Taschuk. Copying a new-style class that had a reference to itself
2866 didn't work. (The same thing worked fine for old-style classes.)
2867 Builtin functions are now treated as atomic, fixing bug #746304.
2869 - difflib.py has two new functions: context_diff() and unified_diff().
2871 - More fixes to urllib (SF 549151): (a) When redirecting, always use
2872 GET. This is common practice and more-or-less sanctioned by the
2873 HTTP standard. (b) Add a handler for 307 redirection, which becomes
2874 an error for POST, but a regular redirect for GET and HEAD
2876 - Added optional 'onerror' argument to os.walk(), to control error
2879 - inspect.is{method|data}descriptor was added, to allow pydoc display
2880 __doc__ of data descriptors.
2882 - Fixed socket speed loss caused by use of the _socketobject wrapper class
2885 - timeit.py now checks the current directory for imports.
2887 - urllib2.py now knows how to order proxy classes, so the user doesn't
2888 have to insert it in front of other classes, nor do dirty tricks like
2889 inserting a "dummy" HTTPHandler after a ProxyHandler when building an
2890 opener with proxy support.
2892 - Iterators have been added for dbm keys.
2894 - random.Random objects can now be pickled.
2899 - pydoc now offers help on keywords and topics.
2901 - Tools/idle is gone; long live Lib/idlelib.
2903 - diff.py prints file diffs in context, unified, or ndiff formats,
2904 providing a command line interface to difflib.py.
2906 - texcheck.py is a new script for making a rough validation of Python LaTeX
2912 - Setting DESTDIR during 'make install' now allows specifying a
2913 different root directory.
2918 - PyType_Ready(): If a type declares that it participates in gc
2919 (Py_TPFLAGS_HAVE_GC), and its base class does not, and its base class's
2920 tp_free slot is the default _PyObject_Del, and type does not define
2921 a tp_free slot itself, _PyObject_GC_Del is assigned to type->tp_free.
2922 Previously _PyObject_Del was inherited, which could at best lead to a
2923 segfault. In addition, if even after this magic the type's tp_free
2924 slot is _PyObject_Del or NULL, and the type is a base type
2925 (Py_TPFLAGS_BASETYPE), TypeError is raised: since the type is a base
2926 type, its dealloc function must call type->tp_free, and since the type
2927 is gc'able, tp_free must not be NULL or _PyObject_Del.
2929 - PyThreadState_SetAsyncExc(): A new API (deliberately accessible only
2930 from C) to interrupt a thread by sending it an exception. It is
2931 intentional that you have to write your own C extension to call it
2943 - test_imp rewritten so that it doesn't raise RuntimeError if run as a
2944 side effect of being imported ("import test.autotest").
2949 - The Windows installer ships with Tcl/Tk 8.4.3 (upgraded from 8.4.1).
2951 - The installer always suggested that Python be installed on the C:
2952 drive, due to a hardcoded "C:" generated by the Wise installation
2953 wizard. People with machines where C: is not the system drive
2954 usually want Python installed on whichever drive is their system drive
2955 instead. We removed the hardcoded "C:", and two testers on machines
2956 where C: is not the system drive report that the installer now
2957 suggests their system drive. Note that you can always select the
2958 directory you want in the "Select Destination Directory" dialog --
2959 that's what it's for.
2964 - There's a new module called "autoGIL", which offers a mechanism to
2965 automatically release the Global Interpreter Lock when an event loop
2966 goes to sleep, allowing other threads to run. It's currently only
2967 supported on OSX, in the Mach-O version.
2968 - The OSA modules now allow direct access to properties of the
2969 toplevel application class (in AppleScript terminology).
2970 - The Package Manager can now update itself.
2972 SourceForge Bugs and Patches Applied
2973 ------------------------------------
2975 430160, 471893, 501716, 542562, 549151, 569574, 595837, 596434,
2976 598163, 604210, 604716, 610332, 612627, 614770, 620190, 621891,
2977 622042, 639139, 640236, 644345, 649742, 649742, 658233, 660022,
2978 661318, 661676, 662807, 662923, 666219, 672855, 678325, 682347,
2979 683486, 684981, 685773, 686254, 692776, 692959, 693094, 696777,
2980 697989, 700827, 703666, 708495, 708604, 708901, 710733, 711902,
2981 713722, 715782, 718286, 719359, 719367, 723136, 723831, 723962,
2982 724588, 724767, 724767, 725942, 726150, 726446, 726869, 727051,
2983 727719, 727719, 727805, 728277, 728563, 728656, 729096, 729103,
2984 729293, 729297, 729300, 729317, 729395, 729622, 729817, 730170,
2985 730296, 730594, 730685, 730826, 730963, 731209, 731403, 731504,
2986 731514, 731626, 731635, 731643, 731644, 731644, 731689, 732124,
2987 732143, 732234, 732284, 732284, 732479, 732761, 732783, 732951,
2988 733667, 733781, 734118, 734231, 734869, 735051, 735293, 735527,
2989 735613, 735694, 736962, 736962, 737970, 738066, 739313, 740055,
2990 740234, 740301, 741806, 742126, 742741, 742860, 742860, 742911,
2991 744041, 744104, 744238, 744687, 744877, 745055, 745478, 745525,
2992 745620, 746012, 746304, 746366, 746801, 746953, 747348, 747667,
2993 747954, 748846, 748849, 748973, 748975, 749191, 749210, 749759,
2994 749831, 749911, 750008, 750092, 750542, 750595, 751038, 751107,
2995 751276, 751451, 751916, 751941, 751956, 751998, 752671, 753451,
2996 753602, 753617, 753845, 753925, 754014, 754340, 754447, 755031,
2997 755087, 755147, 755245, 755683, 755987, 756032, 756996, 757058,
2998 757229, 757818, 757821, 757822, 758112, 758910, 759227, 759889,
2999 760257, 760703, 760792, 761104, 761337, 761519, 761830, 762455
3002 What's New in Python 2.3 beta 1?
3003 ================================
3005 *Release date: 25-Apr-2003*
3010 - New format codes B, H, I, k and K have been implemented for
3011 PyArg_ParseTuple and PyBuild_Value.
3013 - New builtin function sum(seq, start=0) returns the sum of all the
3014 items in iterable object seq, plus start (items are normally numbers,
3015 and cannot be strings).
3017 - bool() called without arguments now returns False rather than
3018 raising an exception. This is consistent with calling the
3019 constructors for the other builtin types -- called without argument
3020 they all return the false value of that type. (SF patch #724135)
3022 - In support of PEP 269 (making the pgen parser generator accessible
3023 from Python), some changes to the pgen code structure were made; a
3024 few files that used to be linked only with pgen are now linked with
3027 - The repr() of a weakref object now shows the __name__ attribute of
3028 the referenced object, if it has one.
3030 - super() no longer ignores data descriptors, except __class__. See
3031 the thread started at
3032 http://mail.python.org/pipermail/python-dev/2003-April/034338.html
3034 - list.insert(i, x) now interprets negative i as it would be
3035 interpreted by slicing, so negative values count from the end of the
3036 list. This was the only place where such an interpretation was not
3037 placed on a list index.
3039 - range() now works even if the arguments are longs with magnitude
3040 larger than sys.maxint, as long as the total length of the sequence
3041 fits. E.g., range(2**100, 2**101, 2**100) is the following list:
3042 [1267650600228229401496703205376L]. (SF patch #707427.)
3044 - Some horridly obscure problems were fixed involving interaction
3045 between garbage collection and old-style classes with "ambitious"
3046 getattr hooks. If an old-style instance didn't have a __del__ method,
3047 but did have a __getattr__ hook, and the instance became reachable
3048 only from an unreachable cycle, and the hook resurrected or deleted
3049 unreachable objects when asked to resolve "__del__", anything up to
3050 a segfault could happen. That's been repaired.
3052 - dict.pop now takes an optional argument specifying a default
3053 value to return if the key is not in the dict. If a default is not
3054 given and the key is not found, a KeyError will still be raised.
3055 Parallel changes were made to UserDict.UserDict and UserDict.DictMixin.
3056 [SF patch #693753] (contributed by Michael Stone.)
3058 - sys.getfilesystemencoding() was added to expose
3059 Py_FileSystemDefaultEncoding.
3061 - New function sys.exc_clear() clears the current exception. This is
3062 rarely needed, but can sometimes be useful to release objects
3063 referenced by the traceback held in sys.exc_info()[2]. (SF patch
3066 - On 64-bit systems, a dictionary could contain duplicate long/int keys
3067 if the key value was larger than 2**32. See SF bug #689659.
3069 - Fixed SF bug #663074. The codec system was using global static
3070 variables to store internal data. As a result, any attempts to use the
3071 unicode system with multiple active interpreters, or successive
3072 interpreter executions, would fail.
3074 - "%c" % u"a" now returns a unicode string instead of raising a
3075 TypeError. u"%c" % 0xffffffff now raises a OverflowError instead
3076 of a ValueError to be consistent with "%c" % 256. See SF patch #710127.
3081 - The socket module now provides the functions inet_pton and inet_ntop
3082 for converting between string and packed representation of IP
3083 addresses. There is also a new module variable, has_ipv6, which is
3084 True iff the current Python has IPv6 support. See SF patch #658327.
3086 - Tkinter wrappers around Tcl variables now pass objects directly
3087 to Tcl, instead of first converting them to strings.
3089 - The .*? pattern in the re module is now special-cased to avoid the
3090 recursion limit. (SF patch #720991 -- many thanks to Gary Herron
3093 - New function sys.call_tracing() allows pdb to debug code
3096 - New function gc.get_referents(obj) returns a list of objects
3097 directly referenced by obj. In effect, it exposes what the object's
3098 tp_traverse slot does, and can be helpful when debugging memory
3101 - The iconv module has been removed from this release.
3103 - The platform-independent routines for packing floats in IEEE formats
3104 (struct.pack's <f, >f, <d, and >d codes; pickle and cPickle's protocol 1
3105 pickling of floats) ignored that rounding can cause a carry to
3106 propagate. The worst consequence was that, in rare cases, <f and >f
3107 could produce strings that, when unpacked again, were a factor of 2
3108 away from the original float. This has been fixed. See SF bug
3111 - New function time.tzset() provides access to the C library tzset()
3112 function, if supported. (SF patch #675422.)
3114 - Using createfilehandler, deletefilehandler, createtimerhandler functions
3115 on Tkinter.tkinter (_tkinter module) no longer crashes the interpreter.
3118 - Modified the fcntl.ioctl() function to allow modification of a passed
3119 mutable buffer (for details see the reference documentation).
3121 - Made user requested changes to the itertools module.
3122 Subsumed the times() function into repeat().
3123 Added chain() and cycle().
3125 - The rotor module is now deprecated; the encryption algorithm it uses
3126 is not believed to be secure, and including crypto code with Python
3127 has implications for exporting and importing it in various countries.
3129 - The socket module now always uses the _socketobject wrapper class, even on
3130 platforms which have dup(2). The makefile() method is built directly
3131 on top of the socket without duplicating the file descriptor, allowing
3132 timeouts to work properly.
3137 - New generator function os.walk() is an easy-to-use alternative to
3138 os.path.walk(). See os module docs for details. os.path.walk()
3139 isn't deprecated at this time, but may become deprecated in a
3142 - Added new module "platform" which provides a wide range of tools
3143 for querying platform dependent features.
3145 - netrc now allows ASCII punctuation characters in passwords.
3147 - shelve now supports the optional writeback argument, and exposes
3148 pickle protocol versions.
3150 - Several methods of nntplib.NNTP have grown an optional file argument
3151 which specifies a file where to divert the command's output
3152 (already supported by the body() method). (SF patch #720468)
3154 - The self-documenting XML server library DocXMLRPCServer was added.
3156 - Support for internationalized domain names has been added through
3157 the 'idna' and 'punycode' encodings, the 'stringprep' module, the
3158 'mkstringprep' tool, and enhancements to the socket and httplib
3161 - htmlentitydefs has two new dictionaries: name2codepoint maps
3162 HTML entity names to Unicode codepoints (as integers).
3163 codepoint2name is the reverse mapping. See SF patch #722017.
3165 - pdb has a new command, "debug", which lets you step through
3166 arbitrary code from the debugger's (pdb) prompt.
3168 - unittest.failUnlessEqual and its equivalent unittest.assertEqual now
3169 return 'not a == b' rather than 'a != b'. This gives the desired
3170 result for classes that define __eq__ without defining __ne__.
3172 - sgmllib now supports SGML marked sections, in particular the
3173 MS Office extensions.
3175 - The urllib module now offers support for the iterator protocol.
3176 SF patch 698520 contributed by Brett Cannon.
3178 - New module timeit provides a simple framework for timing the
3179 execution speed of expressions and statements.
3181 - sets.Set objects now support mixed-type __eq__ and __ne__, instead
3182 of raising TypeError. If x is a Set object and y is a non-Set object,
3183 x == y is False, and x != y is True. This is akin to the change made
3184 for mixed-type comparisons of datetime objects in 2.3a2; more info
3185 about the rationale is in the NEWS entry for that. See also SF bug
3186 report <http://www.python.org/sf/693121>.
3188 - On Unix platforms, if os.listdir() is called with a Unicode argument,
3189 it now returns Unicode strings. (This behavior was added earlier
3190 to the Windows NT/2k/XP version of os.listdir().)
3192 - Distutils: both 'py_modules' and 'packages' keywords can now be specified
3193 in core.setup(). Previously you could supply one or the other, but
3194 not both of them. (SF patch #695090 from Bernhard Herzog)
3196 - New csv package makes it easy to read/write CSV files.
3198 - Module shlex has been extended to allow posix-like shell parsings,
3199 including a split() function for easy spliting of quoted strings and
3200 commands. An iterator interface was also implemented.
3205 - New script combinerefs.py helps analyze new PYTHONDUMPREFS output.
3206 See the module docstring for details.
3211 - Fix problem building on OSF1 because the compiler only accepted
3212 preprocessor directives that start in column 1. (SF bug #691793.)
3217 - Added PyGC_Collect(), equivalent to calling gc.collect().
3219 - PyThreadState_GetDict() was changed not to raise an exception or
3220 issue a fatal error when no current thread state is available. This
3221 makes it possible to print dictionaries when no thread is active.
3223 - LONG_LONG was renamed to PY_LONG_LONG. Extensions that use this and
3224 need compatibility with previous versions can use this:
3226 #ifndef PY_LONG_LONG
3227 #define PY_LONG_LONG LONG_LONG
3230 - Added PyObject_SelfIter() to fill the tp_iter slot for the
3231 typical case where the method returns its self argument.
3233 - The extended type structure used for heap types (new-style
3234 classes defined by Python code using a class statement) is now
3235 exported from object.h as PyHeapTypeObject. (SF patch #696193.)
3245 - test_timeout now requires -u network to be passed to regrtest to run.
3251 - os.fsync() now exists on Windows, and calls the Microsoft _commit()
3254 - New function winsound.MessageBeep() wraps the Win32 API
3260 - os.listdir() now returns Unicode strings on MacOS X when called with
3261 a Unicode argument. See the general news item under "Library".
3263 - A new method MacOS.WMAvailable() returns true if it is safe to access
3264 the window manager, false otherwise.
3266 - EasyDialogs dialogs are now movable-modal, and if the application is
3267 currently in the background they will ask to be moved to the foreground
3270 - OSA Scripting support has improved a lot, and gensuitemodule.py can now
3271 be used by mere mortals. The documentation is now also more or less
3274 - The IDE (in a framework build) now includes introductory documentation
3275 in Apple Help Viewer format.
3278 What's New in Python 2.3 alpha 2?
3279 =================================
3281 *Release date: 19-Feb-2003*
3286 - Negative positions returned from PEP 293 error callbacks are now
3287 treated as being relative to the end of the input string. Positions
3288 that are out of bounds raise an IndexError.
3290 - sys.path[0] (the directory from which the script is loaded) is now
3291 turned into an absolute pathname, unless it is the empty string.
3294 - Finally fixed the bug in compile() and exec where a string ending
3295 with an indented code block but no newline would raise SyntaxError.
3296 This would have been a four-line change in parsetok.c... Except
3297 codeop.py depends on this behavior, so a compilation flag had to be
3298 invented that causes the tokenizer to revert to the old behavior;
3299 this required extra changes to 2 .h files, 2 .c files, and 2 .py
3300 files. (Fixes SF bug #501622.)
3302 - If a new-style class defines neither __new__ nor __init__, its
3303 constructor would ignore all arguments. This is changed now: the
3304 constructor refuses arguments in this case. This might break code
3305 that worked under Python 2.2. The simplest fix is to add a no-op
3306 __init__: ``def __init__(self, *args, **kw): pass``.
3308 - Through a bytecode optimizer bug (and I bet you didn't even know
3309 Python *had* a bytecode optimizer :-), "unsigned" hex/oct constants
3310 with a leading minus sign would come out with the wrong sign.
3311 ("Unsigned" hex/oct constants are those with a face value in the
3312 range sys.maxint+1 through sys.maxint*2+1, inclusive; these have
3313 always been interpreted as negative numbers through sign folding.)
3314 E.g. 0xffffffff is -1, and -(0xffffffff) is 1, but -0xffffffff would
3315 come out as -4294967295. This was the case in Python 2.2 through
3316 2.2.2 and 2.3a1, and in Python 2.4 it will once again have that
3317 value, but according to PEP 237 it really needs to be 1 now. This
3318 will be backported to Python 2.2.3 a well. (SF #660455)
3320 - int(s, base) sometimes sign-folds hex and oct constants; it only
3321 does this when base is 0 and s.strip() starts with a '0'. When the
3322 sign is actually folded, as in int("0xffffffff", 0) on a 32-bit
3323 machine, which returns -1, a FutureWarning is now issued; in Python
3324 2.4, this will return 4294967295L, as do int("+0xffffffff", 0) and
3325 int("0xffffffff", 16) right now. (PEP 347)
3327 - super(X, x): x may now be a proxy for an X instance, i.e.
3328 issubclass(x.__class__, X) but not issubclass(type(x), X).
3330 - isinstance(x, X): if X is a new-style class, this is now equivalent
3331 to issubclass(type(x), X) or issubclass(x.__class__, X). Previously
3332 only type(x) was tested. (For classic classes this was already the
3335 - compile(), eval() and the exec statement now fully support source code
3336 passed as unicode strings.
3338 - int subclasses can be initialized with longs if the value fits in an int.
3341 - long(string, base) takes time linear in len(string) when base is a power
3342 of 2 now. It used to take time quadratic in len(string).
3344 - filter returns now Unicode results for Unicode arguments.
3346 - raw_input can now return Unicode objects.
3348 - List objects' sort() method now accepts None as the comparison function.
3349 Passing None is semantically identical to calling sort() with no
3352 - Fixed crash when printing a subclass of str and __str__ returned self.
3355 - Fixed an invalid RuntimeWarning and an undetected error when trying
3356 to convert a long integer into a float which couldn't fit.
3359 - Function objects now have a __module__ attribute that is bound to
3360 the name of the module in which the function was defined. This
3361 applies for C functions and methods as well as functions and methods
3362 defined in Python. This attribute is used by pickle.whichmodule(),
3363 which changes the behavior of whichmodule slightly. In Python 2.2
3364 whichmodule() returns "__main__" for functions that are not defined
3365 at the top-level of a module (examples: methods, nested functions).
3366 Now whichmodule() will return the proper module name.
3371 - operator.isNumberType() now checks that the object has a nb_int or
3372 nb_float slot, rather than simply checking whether it has a non-NULL
3373 tp_as_number pointer.
3375 - The imp module now has ways to acquire and release the "import
3376 lock": imp.acquire_lock() and imp.release_lock(). Note: this is a
3377 reentrant lock, so releasing the lock only truly releases it when
3378 this is the last release_lock() call. You can check with
3379 imp.lock_held(). (SF bug #580952 and patch #683257.)
3381 - Change to cPickle to match pickle.py (see below and PEP 307).
3383 - Fix some bugs in the parser module. SF bug #678518.
3385 - Thanks to Scott David Daniels, a subtle bug in how the zlib
3386 extension implemented flush() was fixed. Scott also rewrote the
3387 zlib test suite using the unittest module. (SF bug #640230 and
3390 - Added an itertools module containing high speed, memory efficient
3391 looping constructs inspired by tools from Haskell and SML.
3393 - The SSL module now handles sockets with a timeout set correctly (SF
3394 patch #675750, fixing SF bug #675552).
3396 - os/posixmodule has grown the sysexits.h constants (EX_OK and friends).
3398 - Fixed broken threadstate swap in readline that could cause fatal
3399 errors when a readline hook was being invoked while a background
3400 thread was active. (SF bugs #660476 and #513033.)
3402 - fcntl now exposes the strops.h I_* constants.
3404 - Fix a crash on Solaris that occurred when calling close() on
3405 an mmap'ed file which was already closed. (SF patch #665913)
3407 - Fixed several serious bugs in the zipimport implementation.
3411 The date class is now properly subclassable. (SF bug #720908)
3413 The datetime and datetimetz classes have been collapsed into a single
3414 datetime class, and likewise the time and timetz classes into a single
3415 time class. Previously, a datetimetz object with tzinfo=None acted
3416 exactly like a datetime object, and similarly for timetz. This wasn't
3417 enough of a difference to justify distinct classes, and life is simpler
3420 today() and now() now round system timestamps to the closest
3421 microsecond <http://www.python.org/sf/661086>. This repairs an
3422 irritation most likely seen on Windows systems.
3424 In dt.astimezone(tz), if tz.utcoffset(dt) returns a duration,
3425 ValueError is raised if tz.dst(dt) returns None (2.3a1 treated it
3426 as 0 instead, but a tzinfo subclass wishing to participate in
3427 time zone conversion has to take a stand on whether it supports
3428 DST; if you don't care about DST, then code dst() to return 0 minutes,
3429 meaning that DST is never in effect).
3431 The tzinfo methods utcoffset() and dst() must return a timedelta object
3432 (or None) now. In 2.3a1 they could also return an int or long, but that
3433 was an unhelpfully redundant leftover from an earlier version wherein
3434 they couldn't return a timedelta. TOOWTDI.
3436 The example tzinfo class for local time had a bug. It was replaced
3437 by a later example coded by Guido.
3439 datetime.astimezone(tz) no longer raises an exception when the
3440 input datetime has no UTC equivalent in tz. For typical "hybrid" time
3441 zones (a single tzinfo subclass modeling both standard and daylight
3442 time), this case can arise one hour per year, at the hour daylight time
3443 ends. See new docs for details. In short, the new behavior mimics
3444 the local wall clock's behavior of repeating an hour in local time.
3446 dt.astimezone() can no longer be used to convert between naive and aware
3447 datetime objects. If you merely want to attach, or remove, a tzinfo
3448 object, without any conversion of date and time members, use
3449 dt.replace(tzinfo=whatever) instead, where "whatever" is None or a
3450 tzinfo subclass instance.
3452 A new method tzinfo.fromutc(dt) can be overridden in tzinfo subclasses
3453 to give complete control over how a UTC time is to be converted to
3454 a local time. The default astimezone() implementation calls fromutc()
3455 as its last step, so a tzinfo subclass can affect that too by overriding
3456 fromutc(). It's expected that the default fromutc() implementation will
3457 be suitable as-is for "almost all" time zone subclasses, but the
3458 creativity of political time zone fiddling appears unbounded -- fromutc()
3459 allows the highly motivated to emulate any scheme expressible in Python.
3461 datetime.now(): The optional tzinfo argument was undocumented (that's
3462 repaired), and its name was changed to tz ("tzinfo" is overloaded enough
3463 already). With a tz argument, now(tz) used to return the local date
3464 and time, and attach tz to it, without any conversion of date and time
3465 members. This was less than useful. Now now(tz) returns the current
3466 date and time as local time in tz's time zone, akin to ::
3468 tz.fromutc(datetime.utcnow().replace(tzinfo=utc))
3470 where "utc" is an instance of a tzinfo subclass modeling UTC. Without
3471 a tz argument, now() continues to return the current local date and time,
3472 as a naive datetime object.
3474 datetime.fromtimestamp(): Like datetime.now() above, this had less than
3475 useful behavior when the optional tinzo argument was specified. See
3476 also SF bug report <http://www.python.org/sf/660872>.
3478 date and datetime comparison: In order to prevent comparison from
3479 falling back to the default compare-object-addresses strategy, these
3480 raised TypeError whenever they didn't understand the other object type.
3481 They still do, except when the other object has a "timetuple" attribute,
3482 in which case they return NotImplemented now. This gives other
3483 datetime objects (e.g., mxDateTime) a chance to intercept the
3486 date, time, datetime and timedelta comparison: When the exception
3487 for mixed-type comparisons in the last paragraph doesn't apply, if
3488 the comparison is == then False is returned, and if the comparison is
3489 != then True is returned. Because dict lookup and the "in" operator
3490 only invoke __eq__, this allows, for example, ::
3492 if some_datetime in some_sequence:
3496 some_dict[some_timedelta] = whatever
3498 to work as expected, without raising TypeError just because the
3499 sequence is heterogeneous, or the dict has mixed-type keys. [This
3500 seems like a good idea to implement for all mixed-type comparisons
3501 that don't want to allow falling back to address comparison.]
3503 The constructors building a datetime from a timestamp could raise
3504 ValueError if the platform C localtime()/gmtime() inserted "leap
3505 seconds". Leap seconds are ignored now. On such platforms, it's
3506 possible to have timestamps that differ by a second, yet where
3507 datetimes constructed from them are equal.
3509 The pickle format of date, time and datetime objects has changed
3510 completely. The undocumented pickler and unpickler functions no
3511 longer exist. The undocumented __setstate__() and __getstate__()
3512 methods no longer exist either.
3517 - The logging module was updated slightly; the WARN level was renamed
3518 to WARNING, and the matching function/method warn() to warning().
3520 - The pickle and cPickle modules were updated with a new pickling
3521 protocol (documented by pickletools.py, see below) and several
3522 extensions to the pickle customization API (__reduce__, __setstate__
3523 etc.). The copy module now uses more of the pickle customization
3524 API to copy objects that don't implement __copy__ or __deepcopy__.
3525 See PEP 307 for details.
3527 - The distutils "register" command now uses http://www.python.org/pypi
3528 as the default repository. (See PEP 301.)
3530 - the platform dependent path related variables sep, altsep, extsep,
3531 pathsep, curdir, pardir and defpath are now defined in the platform
3532 dependent path modules (e.g. ntpath.py) rather than os.py, so these
3533 variables are now available via os.path. They continue to be
3534 available from the os module.
3535 (see <http://www.python.org/sf/680789>).
3537 - array.array was added to the types repr.py knows about (see
3538 <http://www.python.org/sf/680789>).
3540 - The new pickletools.py contains lots of documentation about pickle
3541 internals, and supplies some helpers for working with pickles, such as
3542 a symbolic pickle disassembler.
3544 - Xmlrpclib.py now supports the builtin boolean type.
3546 - py_compile has a new 'doraise' flag and a new PyCompileError
3549 - SimpleXMLRPCServer now supports CGI through the CGIXMLRPCRequestHandler
3552 - The sets module now raises TypeError in __cmp__, to clarify that
3553 sets are not intended to be three-way-compared; the comparison
3554 operators are overloaded as subset/superset tests.
3556 - Bastion.py and rexec.py are disabled. These modules are not safe in
3559 - realpath is now exported when doing ``from poxixpath import *``.
3560 It is also exported for ntpath, macpath, and os2emxpath.
3563 - New module tarfile from Lars Gustäbel provides a comprehensive interface
3564 to tar archive files with transparent gzip and bzip2 compression.
3565 See SF patch #651082.
3567 - urlparse can now parse imap:// URLs. See SF feature request #618024.
3569 - Tkinter.Canvas.scan_dragto() provides an optional parameter to support
3570 the gain value which is passed to Tk. SF bug# 602259.
3572 - Fix logging.handlers.SysLogHandler protocol when using UNIX domain sockets.
3573 See SF patch #642974.
3575 - The dospath module was deleted. Use the ntpath module when manipulating
3576 DOS paths from other platforms.
3581 - Two new scripts (db2pickle.py and pickle2db.py) were added to the
3582 Tools/scripts directory to facilitate conversion from the old bsddb module
3583 to the new one. While the user-visible API of the new module is
3584 compatible with the old one, it's likely that the version of the
3585 underlying database library has changed. To convert from the old library,
3586 run the db2pickle.py script using the old version of Python to convert it
3587 to a pickle file. After upgrading Python, run the pickle2db.py script
3588 using the new version of Python to reconstitute your database. For
3591 % python2.2 db2pickle.py -h some.db > some.pickle
3592 % python2.3 pickle2db.py -h some.db.new < some.pickle
3594 Run the scripts without any args to get a usage message.
3600 - The audio driver tests (test_ossaudiodev.py and
3601 test_linuxaudiodev.py) are no longer run by default. This is
3602 because they don't always work, depending on your hardware and
3603 software. To run these tests, you must use an invocation like ::
3605 ./python Lib/test/regrtest.py -u audio test_ossaudiodev
3607 - On systems which build using the configure script, compiler flags which
3608 used to be lumped together using the OPT flag have been split into two
3609 groups, OPT and BASECFLAGS. OPT is meant to carry just optimization- and
3610 debug-related flags like "-g" and "-O3". BASECFLAGS is meant to carry
3611 compiler flags that are required to get a clean compile. On some
3612 platforms (many Linux flavors in particular) BASECFLAGS will be empty by
3613 default. On others, such as Mac OS X and SCO, it will contain required
3614 flags. This change allows people building Python to override OPT without
3615 fear of clobbering compiler flags which are required to get a clean build.
3617 - On Darwin/Mac OS X platforms, /sw/lib and /sw/include are added to the
3618 relevant search lists in setup.py. This allows users building Python to
3619 take advantage of the many packages available from the fink project
3620 <http://fink.sf.net/>.
3622 - A new Makefile target, scriptsinstall, installs a number of useful scripts
3623 from the Tools/scripts directory.
3628 - PyEval_GetFrame() is now declared to return a ``PyFrameObject *``
3629 instead of a plain ``PyObject *``. (SF patch #686601.)
3631 - PyNumber_Check() now checks that the object has a nb_int or nb_float
3632 slot, rather than simply checking whether it has a non-NULL
3633 tp_as_number pointer.
3635 - A C type that inherits from a base type that defines tp_as_buffer
3636 will now inherit the tp_as_buffer pointer if it doesn't define one.
3639 - The PyArg_Parse functions now issue a DeprecationWarning if a float
3640 argument is provided when an integer is specified (this affects the 'b',
3641 'B', 'h', 'H', 'i', and 'l' codes). Future versions of Python will
3647 - Several tests weren't being run from regrtest.py (test_timeout.py,
3648 test_tarfile.py, test_netrc.py, test_multifile.py,
3649 test_importhooks.py and test_imp.py). Now they are. (Note to
3650 developers: please read Lib/test/README when creating a new test, to
3651 make sure to do it right! All tests need to use either unittest or
3654 - Added test_posix.py, a test suite for the posix module.
3656 - Added test_hexoct.py, a test suite for hex/oct constant folding.
3661 - The timeout code for socket connect() didn't work right; this has
3662 now been fixed. test_timeout.py should pass (at least most of the
3665 - distutils' msvccompiler class now passes the preprocessor options to
3666 the resource compiler. See SF patch #669198.
3668 - The bsddb module now ships with Sleepycat's 4.1.25.NC, the latest
3669 release without strong cryptography.
3671 - sys.path[0], if it contains a directory name, is now always an
3672 absolute pathname. (SF patch #664376.)
3674 - The new logging package is now installed by the Windows installer. It
3675 wasn't in 2.3a1 due to oversight.
3680 - There are new dialogs EasyDialogs.AskFileForOpen, AskFileForSave
3681 and AskFolder. The old macfs.StandardGetFile and friends are deprecated.
3683 - Most of the standard library now uses pathnames or FSRefs in preference
3684 of FSSpecs, and use the underlying Carbon.File and Carbon.Folder modules
3685 in stead of macfs. macfs will probably be deprecated in the future.
3687 - Type Carbon.File.FSCatalogInfo and supporting methods have been implemented.
3688 This also makes macfs.FSSpec.SetDates() work again.
3690 - There is a new module pimp, the package install manager for Python, and
3691 accompanying applet PackageManager. These allow you to easily download
3692 and install pretested extension packages either in source or binary
3693 form. Only in MacPython-OSX.
3695 - Applets are now built with bundlebuilder in MacPython-OSX, which should make
3696 them more robust and also provides a path towards BuildApplication. The
3697 downside of this change is that applets can no longer be run from the
3698 Terminal window, this will hopefully be fixed in the 2.3b1.
3701 What's New in Python 2.3 alpha 1?
3702 =================================
3704 *Release date: 31-Dec-2002*
3706 Type/class unification and new-style classes
3707 --------------------------------------------
3709 - One can now assign to __bases__ and __name__ of new-style classes.
3711 - dict() now accepts keyword arguments so that dict(one=1, two=2)
3712 is the equivalent of {"one": 1, "two": 2}. Accordingly,
3713 the existing (but undocumented) 'items' keyword argument has
3714 been eliminated. This means that dict(items=someMapping) now has
3715 a different meaning than before.
3717 - int() now returns a long object if the argument is outside the
3718 integer range, so int("4" * 1000), int(1e200) and int(1L<<1000) will
3719 all return long objects instead of raising an OverflowError.
3721 - Assignment to __class__ is disallowed if either the old or the new
3722 class is a statically allocated type object (such as defined by an
3723 extension module). This prevents anomalies like 2.__class__ = bool.
3725 - New-style object creation and deallocation have been sped up
3726 significantly; they are now faster than classic instance creation
3729 - The __slots__ variable can now mention "private" names, and the
3730 right thing will happen (e.g. __slots__ = ["__foo"]).
3732 - The built-ins slice() and buffer() are now callable types. The
3733 types classobj (formerly class), code, function, instance, and
3734 instancemethod (formerly instance-method), which have no built-in
3735 names but are accessible through the types module, are now also
3736 callable. The type dict-proxy is renamed to dictproxy.
3738 - Cycles going through the __class__ link of a new-style instance are
3739 now detected by the garbage collector.
3741 - Classes using __slots__ are now properly garbage collected.
3744 - Tightened the __slots__ rules: a slot name must be a valid Python
3747 - The constructor for the module type now requires a name argument and
3748 takes an optional docstring argument. Previously, this constructor
3749 ignored its arguments. As a consequence, deriving a class from a
3750 module (not from the module type) is now illegal; previously this
3751 created an unnamed module, just like invoking the module type did.
3754 - A new type object, 'basestring', is added. This is a common base type
3755 for 'str' and 'unicode', and can be used instead of
3756 types.StringTypes, e.g. to test whether something is "a string":
3757 isinstance(x, basestring) is True for Unicode and 8-bit strings. This
3758 is an abstract base class and cannot be instantiated directly.
3760 - Changed new-style class instantiation so that when C's __new__
3761 method returns something that's not a C instance, its __init__ is
3762 not called. [SF bug #537450]
3764 - Fixed super() to work correctly with class methods. [SF bug #535444]
3766 - If you try to pickle an instance of a class that has __slots__ but
3767 doesn't define or override __getstate__, a TypeError is now raised.
3768 This is done by adding a bozo __getstate__ to the class that always
3769 raises TypeError. (Before, this would appear to be pickled, but the
3770 state of the slots would be lost.)
3775 - Import from zipfiles is now supported. The name of a zipfile placed
3776 on sys.path causes the import statement to look for importable Python
3777 modules (with .py, pyc and .pyo extensions) and packages inside the
3778 zipfile. The zipfile import follows the specification (though not
3779 the sample implementation) of PEP 273. The semantics of __path__ are
3780 compatible with those that have been implemented in Jython since
3783 - PEP 302 has been accepted. Although it was initially developed to
3784 support zipimport, it offers a new, general import hook mechanism.
3785 Several new variables have been added to the sys module:
3786 sys.meta_path, sys.path_hooks, and sys.path_importer_cache; these
3787 make extending the import statement much more convenient than
3788 overriding the __import__ built-in function. For a description of
3791 - A frame object's f_lineno attribute can now be written to from a
3792 trace function to change which line will execute next. A command to
3793 exploit this from pdb has been added. [SF patch #643835]
3795 - The _codecs support module for codecs.py was turned into a builtin
3796 module to assure that at least the builtin codecs are available
3797 to the Python parser for source code decoding according to PEP 263.
3799 - issubclass now supports a tuple as the second argument, just like
3800 isinstance does. ``issubclass(X, (A, B))`` is equivalent to
3801 ``issubclass(X, A) or issubclass(X, B)``.
3803 - Thanks to Armin Rigo, the last known way to provoke a system crash
3804 by cleverly arranging for a comparison function to mutate a list
3805 during a list.sort() operation has been fixed. The effect of
3806 attempting to mutate a list, or even to inspect its contents or
3807 length, while a sort is in progress, is not defined by the language.
3808 The C implementation of Python 2.3 attempts to detect mutations,
3809 and raise ValueError if one occurs, but there's no guarantee that
3810 all mutations will be caught, or that any will be caught across
3811 releases or implementations.
3813 - Unicode file name processing for Windows (PEP 277) is implemented.
3814 All platforms now have an os.path.supports_unicode_filenames attribute,
3815 which is set to True on Windows NT/2000/XP, and False elsewhere.
3817 - Codec error handling callbacks (PEP 293) are implemented.
3818 Error handling in unicode.encode or str.decode can now be customized.
3820 - A subtle change to the semantics of the built-in function intern():
3821 interned strings are no longer immortal. You must keep a reference
3822 to the return value intern() around to get the benefit.
3824 - Use of 'None' as a variable, argument or attribute name now
3825 issues a SyntaxWarning. In the future, None may become a keyword.
3827 - SET_LINENO is gone. co_lnotab is now consulted to determine when to
3828 call the trace function. C code that accessed f_lineno should call
3829 PyCode_Addr2Line instead (f_lineno is still there, but only kept up
3830 to date when there is a trace function set).
3832 - There's a new warning category, FutureWarning. This is used to warn
3833 about a number of situations where the value or sign of an integer
3834 result will change in Python 2.4 as a result of PEP 237 (integer
3835 unification). The warnings implement stage B0 mentioned in that
3836 PEP. The warnings are about the following situations:
3838 - Octal and hex literals without 'L' prefix in the inclusive range
3839 [0x80000000..0xffffffff]; these are currently negative ints, but
3840 in Python 2.4 they will be positive longs with the same bit
3843 - Left shifts on integer values that cause the outcome to lose
3844 bits or have a different sign than the left operand. To be
3845 precise: x<<n where this currently doesn't yield the same value
3846 as long(x)<<n; in Python 2.4, the outcome will be long(x)<<n.
3848 - Conversions from ints to string that show negative values as
3849 unsigned ints in the inclusive range [0x80000000..0xffffffff];
3850 this affects the functions hex() and oct(), and the string
3851 formatting codes %u, %o, %x, and %X. In Python 2.4, these will
3852 show signed values (e.g. hex(-1) currently returns "0xffffffff";
3853 in Python 2.4 it will return "-0x1").
3855 - The bits manipulated under the cover by sys.setcheckinterval() have
3856 been changed. Both the check interval and the ticker used to be
3857 per-thread values. They are now just a pair of global variables.
3858 In addition, the default check interval was boosted from 10 to 100
3859 bytecode instructions. This may have some effect on systems that
3860 relied on the old default value. In particular, in multi-threaded
3861 applications which try to be highly responsive, response time will
3862 increase by some (perhaps imperceptible) amount.
3864 - When multiplying very large integers, a version of the so-called
3865 Karatsuba algorithm is now used. This is most effective if the
3866 inputs have roughly the same size. If they both have about N digits,
3867 Karatsuba multiplication has O(N**1.58) runtime (the exponent is
3868 log_base_2(3)) instead of the previous O(N**2). Measured results may
3869 be better or worse than that, depending on platform quirks. Besides
3870 the O() improvement in raw instruction count, the Karatsuba algorithm
3871 appears to have much better cache behavior on extremely large integers
3872 (starting in the ballpark of a million bits). Note that this is a
3873 simple implementation, and there's no intent here to compete with,
3874 e.g., GMP. It gives a very nice speedup when it applies, but a package
3875 devoted to fast large-integer arithmetic should run circles around it.
3877 - u'%c' will now raise a ValueError in case the argument is an
3878 integer outside the valid range of Unicode code point ordinals.
3880 - The tempfile module has been overhauled for enhanced security. The
3881 mktemp() function is now deprecated; new, safe replacements are
3882 mkstemp() (for files) and mkdtemp() (for directories), and the
3883 higher-level functions NamedTemporaryFile() and TemporaryFile().
3884 Use of some global variables in this module is also deprecated; the
3885 new functions have keyword arguments to provide the same
3886 functionality. All Lib, Tools and Demo modules that used the unsafe
3887 interfaces have been updated to use the safe replacements. Thanks
3890 - When x is an object whose class implements __mul__ and __rmul__,
3891 1.0*x would correctly invoke __rmul__, but 1*x would erroneously
3892 invoke __mul__. This was due to the sequence-repeat code in the int
3893 type. This has been fixed now.
3895 - Previously, "str1 in str2" required str1 to be a string of length 1.
3896 This restriction has been relaxed to allow str1 to be a string of
3897 any length. Thus "'el' in 'hello world'" returns True now.
3899 - File objects are now their own iterators. For a file f, iter(f) now
3900 returns f (unless f is closed), and f.next() is similar to
3901 f.readline() when EOF is not reached; however, f.next() uses a
3902 readahead buffer that messes up the file position, so mixing
3903 f.next() and f.readline() (or other methods) doesn't work right.
3904 Calling f.seek() drops the readahead buffer, but other operations
3905 don't. It so happens that this gives a nice additional speed boost
3906 to "for line in file:"; the xreadlines method and corresponding
3907 module are now obsolete. Thanks to Oren Tirosh!
3909 - Encoding declarations (PEP 263, phase 1) have been implemented. A
3910 comment of the form "# -*- coding: <encodingname> -*-" in the first
3911 or second line of a Python source file indicates the encoding.
3913 - list.sort() has a new implementation. While cross-platform results
3914 may vary, and in data-dependent ways, this is much faster on many
3915 kinds of partially ordered lists than the previous implementation,
3916 and reported to be just as fast on randomly ordered lists on
3917 several major platforms. This sort is also stable (if A==B and A
3918 precedes B in the list at the start, A precedes B after the sort too),
3919 although the language definition does not guarantee stability. A
3920 potential drawback is that list.sort() may require temp space of
3921 len(list)*2 bytes (``*4`` on a 64-bit machine). It's therefore possible
3922 for list.sort() to raise MemoryError now, even if a comparison function
3923 does not. See <http://www.python.org/sf/587076> for full details.
3925 - All standard iterators now ensure that, once StopIteration has been
3926 raised, all future calls to next() on the same iterator will also
3927 raise StopIteration. There used to be various counterexamples to
3928 this behavior, which could caused confusion or subtle program
3929 breakage, without any benefits. (Note that this is still an
3930 iterator's responsibility; the iterator framework does not enforce
3933 - Ctrl+C handling on Windows has been made more consistent with
3934 other platforms. KeyboardInterrupt can now reliably be caught,
3935 and Ctrl+C at an interactive prompt no longer terminates the
3936 process under NT/2k/XP (it never did under Win9x). Ctrl+C will
3937 interrupt time.sleep() in the main thread, and any child processes
3938 created via the popen family (on win2k; we can't make win9x work
3939 reliably) are also interrupted (as generally happens on for Linux/Unix.)
3940 [SF bugs 231273, 439992 and 581232]
3942 - sys.getwindowsversion() has been added on Windows. This
3943 returns a tuple with information about the version of Windows
3946 - Slices and repetitions of buffer objects now consistently return
3947 a string. Formerly, strings would be returned most of the time,
3948 but a buffer object would be returned when the repetition count
3949 was one or when the slice range was all inclusive.
3951 - Unicode objects in sys.path are no longer ignored but treated
3954 - Fixed string.startswith and string.endswith builtin methods
3955 so they accept negative indices. [SF bug 493951]
3957 - Fixed a bug with a continue inside a try block and a yield in the
3958 finally clause. [SF bug 567538]
3960 - Most builtin sequences now support "extended slices", i.e. slices
3961 with a third "stride" parameter. For example, "hello world"[::-1]
3962 gives "dlrow olleh".
3964 - A new warning PendingDeprecationWarning was added to provide
3965 direction on features which are in the process of being deprecated.
3966 The warning will not be printed by default. To see the pending
3967 deprecations, use -Walways::PendingDeprecationWarning::
3968 as a command line option or warnings.filterwarnings() in code.
3970 - Deprecated features of xrange objects have been removed as
3971 promised. The start, stop, and step attributes and the tolist()
3972 method no longer exist. xrange repetition and slicing have been
3975 - New builtin function enumerate(x), from PEP 279. Example:
3976 enumerate("abc") is an iterator returning (0,"a"), (1,"b"), (2,"c").
3977 The argument can be an arbitrary iterable object.
3979 - The assert statement no longer tests __debug__ at runtime. This means
3980 that assert statements cannot be disabled by assigning a false value
3983 - A method zfill() was added to str and unicode, that fills a numeric
3984 string to the left with zeros. For example,
3985 "+123".zfill(6) -> "+00123".
3987 - Complex numbers supported divmod() and the // and % operators, but
3988 these make no sense. Since this was documented, they're being
3991 - String and unicode methods lstrip(), rstrip() and strip() now take
3992 an optional argument that specifies the characters to strip. For
3993 example, "Foo!!!?!?!?".rstrip("?!") -> "Foo".
3995 - There's a new dictionary constructor (a class method of the dict
3996 class), dict.fromkeys(iterable, value=None). It constructs a
3997 dictionary with keys taken from the iterable and all values set to a
3998 single value. It can be used for building sets and for removing
3999 duplicates from sequences.
4001 - Added a new dict method pop(key). This removes and returns the
4002 value corresponding to key. [SF patch #539949]
4004 - A new built-in type, bool, has been added, as well as built-in
4005 names for its two values, True and False. Comparisons and sundry
4006 other operations that return a truth value have been changed to
4007 return a bool instead. Read PEP 285 for an explanation of why this
4008 is backward compatible.
4010 - Fixed two bugs reported as SF #535905: under certain conditions,
4011 deallocating a deeply nested structure could cause a segfault in the
4012 garbage collector, due to interaction with the "trashcan" code;
4013 access to the current frame during destruction of a local variable
4014 could access a pointer to freed memory.
4016 - The optional object allocator ("pymalloc") has been enabled by
4017 default. The recommended practice for memory allocation and
4018 deallocation has been streamlined. A header file is included,
4019 Misc/pymemcompat.h, which can be bundled with 3rd party extensions
4020 and lets them use the same API with Python versions from 1.5.2
4023 - PyErr_Display will provide file and line information for all exceptions
4024 that have an attribute print_file_and_line, not just SyntaxErrors.
4026 - The UTF-8 codec will now encode and decode Unicode surrogates
4027 correctly and without raising exceptions for unpaired ones.
4029 - Universal newlines (PEP 278) is implemented. Briefly, using 'U'
4030 instead of 'r' when opening a text file for reading changes the line
4031 ending convention so that any of '\r', '\r\n', and '\n' is
4032 recognized (even mixed in one file); all three are converted to
4033 '\n', the standard Python line end character.
4035 - file.xreadlines() now raises a ValueError if the file is closed:
4036 Previously, an xreadlines object was returned which would raise
4037 a ValueError when the xreadlines.next() method was called.
4039 - sys.exit() inadvertently allowed more than one argument.
4040 An exception will now be raised if more than one argument is used.
4042 - Changed evaluation order of dictionary literals to conform to the
4043 general left to right evaluation order rule. Now {f1(): f2()} will
4046 - Fixed bug #521782: when a file was in non-blocking mode, file.read()
4047 could silently lose data or wrongly throw an unknown error.
4049 - The sq_repeat, sq_inplace_repeat, sq_concat and sq_inplace_concat
4050 slots are now always tried after trying the corresponding nb_* slots.
4051 This fixes a number of minor bugs (see bug #624807).
4053 - Fix problem with dynamic loading on 64-bit AIX (see bug #639945).
4058 - Added three operators to the operator module:
4059 operator.pow(a,b) which is equivalent to: a**b.
4060 operator.is_(a,b) which is equivalent to: a is b.
4061 operator.is_not(a,b) which is equivalent to: a is not b.
4063 - posix.openpty now works on all systems that have /dev/ptmx.
4065 - A module zipimport exists to support importing code from zip
4068 - The new datetime module supplies classes for manipulating dates and
4069 times. The basic design came from the Zope "fishbowl process", and
4070 favors practical commercial applications over calendar esoterica. See
4072 http://www.zope.org/Members/fdrake/DateTimeWiki/FrontPage
4074 - _tkinter now returns Tcl objects, instead of strings. Objects which
4075 have Python equivalents are converted to Python objects, other objects
4076 are wrapped. This can be configured through the wantobjects method,
4077 or Tkinter.wantobjects.
4079 - The PyBSDDB wrapper around the Sleepycat Berkeley DB library has
4080 been added as the package bsddb. The traditional bsddb module is
4081 still available in source code, but not built automatically anymore,
4082 and is now named bsddb185. This supports Berkeley DB versions from
4083 3.0 to 4.1. For help converting your databases from the old module (which
4084 probably used an obsolete version of Berkeley DB) to the new module, see
4085 the db2pickle.py and pickle2db.py scripts described in the Tools/Demos
4088 - unicodedata was updated to Unicode 3.2. It supports normalization
4089 and names for Hangul syllables and CJK unified ideographs.
4091 - resource.getrlimit() now returns longs instead of ints.
4093 - readline now dynamically adjusts its input/output stream if
4094 sys.stdin/stdout changes.
4096 - The _tkinter module (and hence Tkinter) has dropped support for
4097 Tcl/Tk 8.0 and 8.1. Only Tcl/Tk versions 8.2, 8.3 and 8.4 are
4100 - cPickle.BadPickleGet is now a class.
4102 - The time stamps in os.stat_result are floating point numbers
4103 after stat_float_times has been called.
4105 - If the size passed to mmap.mmap() is larger than the length of the
4106 file on non-Windows platforms, a ValueError is raised. [SF bug 585792]
4108 - The xreadlines module is slated for obsolescence.
4110 - The strptime function in the time module is now always available (a
4111 Python implementation is used when the C library doesn't define it).
4113 - The 'new' module is no longer an extension, but a Python module that
4114 only exists for backwards compatibility. Its contents are no longer
4115 functions but callable type objects.
4117 - The bsddb.*open functions can now take 'None' as a filename.
4118 This will create a temporary in-memory bsddb that won't be
4121 - posix.getloadavg, posix.lchown, posix.killpg, posix.mknod, and
4122 posix.getpgid have been added where available.
4124 - The locale module now exposes the C library's gettext interface. It
4125 also has a new function getpreferredencoding.
4127 - A security hole ("double free") was found in zlib-1.1.3, a popular
4128 third party compression library used by some Python modules. The
4129 hole was quickly plugged in zlib-1.1.4, and the Windows build of
4130 Python now ships with zlib-1.1.4.
4132 - pwd, grp, and resource return enhanced tuples now, with symbolic
4135 - array.array is now a type object. A new format character
4136 'u' indicates Py_UNICODE arrays. For those, .tounicode and
4137 .fromunicode methods are available. Arrays now support __iadd__
4140 - dl now builds on every system that has dlfcn.h. Failure in case
4141 of sizeof(int)!=sizeof(long)!=sizeof(void*) is delayed until dl.open
4144 - The sys module acquired a new attribute, api_version, which evaluates
4145 to the value of the PYTHON_API_VERSION macro with which the
4146 interpreter was compiled.
4148 - Fixed bug #470582: sre module would return a tuple (None, 'a', 'ab')
4149 when applying the regular expression '^((a)c)?(ab)$' on 'ab'. It now
4150 returns (None, None, 'ab'), as expected. Also fixed handling of
4151 lastindex/lastgroup match attributes in similar cases. For example,
4152 when running the expression r'(a)(b)?b' over 'ab', lastindex must be
4155 - Fixed bug #581080: sre scanner was not checking the buffer limit
4156 before increasing the current pointer. This was creating an infinite
4157 loop in the search function, once the pointer exceeded the buffer
4160 - The os.fdopen function now enforces a file mode starting with the
4161 letter 'r', 'w' or 'a', otherwise a ValueError is raised. This fixes
4164 - The linuxaudiodev module is now deprecated; it is being replaced by
4165 ossaudiodev. The interface has been extended to cover a lot more of
4166 OSS (see www.opensound.com), including most DSP ioctls and the
4167 OSS mixer API. Documentation forthcoming in 2.3a2.
4172 - imaplib.py now supports SSL (Tino Lange and Piers Lauder).
4174 - Freeze's modulefinder.py has been moved to the standard library;
4175 slightly improved so it will issue less false missing submodule
4176 reports (see sf path #643711 for details). Documentation will follow
4179 - os.path exposes getctime.
4181 - unittest.py now has two additional methods called assertAlmostEqual()
4182 and failIfAlmostEqual(). They implement an approximate comparison
4183 by rounding the difference between the two arguments and comparing
4184 the result to zero. Approximate comparison is essential for
4185 unit tests of floating point results.
4187 - calendar.py now depends on the new datetime module rather than
4188 the time module. As a result, the range of allowable dates
4191 - pdb has a new 'j(ump)' command to select the next line to be
4194 - The distutils created windows installers now can run a
4195 postinstallation script.
4197 - doctest.testmod can now be called without argument, which means to
4198 test the current module.
4200 - When canceling a server that implemented threading with a keyboard
4201 interrupt, the server would shut down but not terminate (waiting on
4202 client threads). A new member variable, daemon_threads, was added to
4203 the ThreadingMixIn class in SocketServer.py to make it explicit that
4204 this behavior needs to be controlled.
4206 - A new module, optparse, provides a fancy alternative to getopt for
4207 command line parsing. It is a slightly modified version of Greg
4208 Ward's Optik package.
4210 - UserDict.py now defines a DictMixin class which defines all dictionary
4211 methods for classes that already have a minimum mapping interface.
4212 This greatly simplifies writing classes that need to be substitutable
4213 for dictionaries (such as the shelve module).
4215 - shelve.py now subclasses from UserDict.DictMixin. Now shelve supports
4216 all dictionary methods. This eases the transition to persistent
4217 storage for scripts originally written with dictionaries in mind.
4219 - shelve.open and the various classes in shelve.py now accept an optional
4220 binary flag, which defaults to False. If True, the values stored in the
4221 shelf are binary pickles.
4223 - A new package, logging, implements the logging API defined by PEP
4224 282. The code is written by Vinay Sajip.
4226 - StreamReader, StreamReaderWriter and StreamRecoder in the codecs
4227 modules are iterators now.
4229 - gzip.py now handles files exceeding 2GB. Files over 4GB also work
4230 now (provided the OS supports it, and Python is configured with large
4231 file support), but in that case the underlying gzip file format can
4232 record only the least-significant 32 bits of the file size, so that
4233 some tools working with gzipped files may report an incorrect file
4236 - xml.sax.saxutils.unescape has been added, to replace entity references
4237 with their entity value.
4239 - Queue.Queue.{put,get} now support an optional timeout argument.
4241 - Various features of Tk 8.4 are exposed in Tkinter.py. The multiple
4242 option of tkFileDialog is exposed as function askopenfile{,name}s.
4244 - Various configure methods of Tkinter have been stream-lined, so that
4245 tag_configure, image_configure, window_configure now return a
4246 dictionary when invoked with no argument.
4248 - Importing the readline module now no longer has the side effect of
4249 calling setlocale(LC_CTYPE, ""). The initial "C" locale, or
4250 whatever locale is explicitly set by the user, is preserved. If you
4251 want repr() of 8-bit strings in your preferred encoding to preserve
4252 all printable characters of that encoding, you have to add the
4253 following code to your $PYTHONSTARTUP file or to your application's
4257 locale.setlocale(locale.LC_CTYPE, "")
4259 - shutil.move was added. shutil.copytree now reports errors as an
4260 exception at the end, instead of printing error messages.
4262 - Encoding name normalization was generalized to not only
4263 replace hyphens with underscores, but also all other non-alphanumeric
4264 characters (with the exception of the dot which is used for Python
4265 package names during lookup). The aliases.py mapping was updated
4266 to the new standard.
4268 - mimetypes has two new functions: guess_all_extensions() which
4269 returns a list of all known extensions for a mime type, and
4270 add_type() which adds one mapping between a mime type and
4271 an extension to the database.
4273 - New module: sets, defines the class Set that implements a mutable
4274 set type using the keys of a dict to represent the set. There's
4275 also a class ImmutableSet which is useful when you need sets of sets
4276 or when you need to use sets as dict keys, and a class BaseSet which
4277 is the base class of the two.
4279 - Added random.sample(population,k) for random sampling without replacement.
4280 Returns a k length list of unique elements chosen from the population.
4282 - random.randrange(-sys.maxint-1, sys.maxint) no longer raises
4283 OverflowError. That is, it now accepts any combination of 'start'
4284 and 'stop' arguments so long as each is in the range of Python's
4287 - Thanks to Raymond Hettinger, random.random() now uses a new core
4288 generator. The Mersenne Twister algorithm is implemented in C,
4289 threadsafe, faster than the previous generator, has an astronomically
4290 large period (2**19937-1), creates random floats to full 53-bit
4291 precision, and may be the most widely tested random number generator
4294 The random.jumpahead(n) method has different semantics for the new
4295 generator. Instead of jumping n steps ahead, it uses n and the
4296 existing state to create a new state. This means that jumpahead()
4297 continues to support multi-threaded code needing generators of
4298 non-overlapping sequences. However, it will break code which relies
4299 on jumpahead moving a specific number of steps forward.
4301 The attributes random.whseed and random.__whseed have no meaning for
4302 the new generator. Code using these attributes should switch to a
4303 new class, random.WichmannHill which is provided for backward
4304 compatibility and to make an alternate generator available.
4306 - New "algorithms" module: heapq, implements a heap queue. Thanks to
4307 Kevin O'Connor for the code and François Pinard for an entertaining
4308 write-up explaining the theory and practical uses of heaps.
4310 - New encoding for the Palm OS character set: palmos.
4312 - binascii.crc32() and the zipfile module had problems on some 64-bit
4313 platforms. These have been fixed. On a platform with 8-byte C longs,
4314 crc32() now returns a signed-extended 4-byte result, so that its value
4315 as a Python int is equal to the value computed a 32-bit platform.
4317 - xml.dom.minidom.toxml and toprettyxml now take an optional encoding
4320 - Some fixes in the copy module: when an object is copied through its
4321 __reduce__ method, there was no check for a __setstate__ method on
4322 the result [SF patch 565085]; deepcopy should treat instances of
4323 custom metaclasses the same way it treats instances of type 'type'
4326 - Sockets now support timeout mode. After s.settimeout(T), where T is
4327 a float expressing seconds, subsequent operations raise an exception
4328 if they cannot be completed within T seconds. To disable timeout
4329 mode, use s.settimeout(None). There's also a module function,
4330 socket.setdefaulttimeout(T), which sets the default for all sockets
4333 - getopt.gnu_getopt was added. This supports GNU-style option
4334 processing, where options can be mixed with non-option arguments.
4336 - Stop using strings for exceptions. String objects used for
4337 exceptions are now classes deriving from Exception. The objects
4338 changed were: Tkinter.TclError, bdb.BdbQuit, macpath.norm_error,
4339 tabnanny.NannyNag, and xdrlib.Error.
4341 - Constants BOM_UTF8, BOM_UTF16, BOM_UTF16_LE, BOM_UTF16_BE,
4342 BOM_UTF32, BOM_UTF32_LE and BOM_UTF32_BE that represent the Byte
4343 Order Mark in UTF-8, UTF-16 and UTF-32 encodings for little and
4344 big endian systems were added to the codecs module. The old names
4345 BOM32_* and BOM64_* were off by a factor of 2.
4347 - Added conversion functions math.degrees() and math.radians().
4349 - math.log() now takes an optional argument: math.log(x[, base]).
4351 - ftplib.retrlines() now tests for callback is None rather than testing
4352 for False. Was causing an error when given a callback object which
4353 was callable but also returned len() as zero. The change may
4354 create new breakage if the caller relied on the undocumented behavior
4355 and called with callback set to [] or some other False value not
4358 - random.gauss() uses a piece of hidden state used by nothing else,
4359 and the .seed() and .whseed() methods failed to reset it. In other
4360 words, setting the seed didn't completely determine the sequence of
4361 results produced by random.gauss(). It does now. Programs repeatedly
4362 mixing calls to a seed method with calls to gauss() may see different
4365 - The pickle.Pickler class grew a clear_memo() method to mimic that
4366 provided by cPickle.Pickler.
4368 - difflib's SequenceMatcher class now does a dynamic analysis of
4369 which elements are so frequent as to constitute noise. For
4370 comparing files as sequences of lines, this generally works better
4371 than the IS_LINE_JUNK function, and function ndiff's linejunk
4372 argument defaults to None now as a result. A happy benefit is
4373 that SequenceMatcher may run much faster now when applied
4374 to large files with many duplicate lines (for example, C program
4375 text with lots of repeated "}" and "return NULL;" lines).
4377 - New Text.dump() method in Tkinter module.
4379 - New distutils commands for building packagers were added to
4380 support pkgtool on Solaris and swinstall on HP-UX.
4382 - distutils now has a new abstract binary packager base class
4383 command/bdist_packager, which simplifies writing packagers.
4384 This will hopefully provide the missing bits to encourage
4385 people to submit more packagers, e.g. for Debian, FreeBSD
4388 - The UTF-16, -LE and -BE stream readers now raise a
4389 NotImplementedError for all calls to .readline(). Previously, they
4390 used to just produce garbage or fail with an encoding error --
4391 UTF-16 is a 2-byte encoding and the C lib's line reading APIs don't
4392 work well with these.
4394 - compileall now supports quiet operation.
4396 - The BaseHTTPServer now implements optional HTTP/1.1 persistent
4399 - socket module: the SSL support was broken out of the main
4400 _socket module C helper and placed into a new _ssl helper
4401 which now gets imported by socket.py if available and working.
4403 - encodings package: added aliases for all supported IANA character
4406 - ftplib: to safeguard the user's privacy, anonymous login will use
4407 "anonymous@" as default password, rather than the real user and host
4410 - webbrowser: tightened up the command passed to os.system() so that
4411 arbitrary shell code can't be executed because a bogus URL was
4414 - gettext.translation has an optional fallback argument, and
4415 gettext.find an optional all argument. Translations will now fallback
4416 on a per-message basis. The module supports plural forms, by means
4417 of gettext.[d]ngettext and Translation.[u]ngettext.
4419 - distutils bdist commands now offer a --skip-build option.
4421 - warnings.warn now accepts a Warning instance as first argument.
4423 - The xml.sax.expatreader.ExpatParser class will no longer create
4424 circular references by using itself as the locator that gets passed
4425 to the content handler implementation. [SF bug #535474]
4427 - The email.Parser.Parser class now properly parses strings regardless
4428 of their line endings, which can be any of \r, \n, or \r\n (CR, LF,
4429 or CRLF). Also, the Header class's constructor default arguments
4430 has changed slightly so that an explicit maxlinelen value is always
4431 honored, and so unicode conversion error handling can be specified.
4433 - distutils' build_ext command now links C++ extensions with the C++
4434 compiler available in the Makefile or CXX environment variable, if
4435 running under \*nix.
4437 - New module bz2: provides a comprehensive interface for the bz2 compression
4438 library. It implements a complete file interface, one-shot (de)compression
4439 functions, and types for sequential (de)compression.
4441 - New pdb command 'pp' which is like 'p' except that it pretty-prints
4442 the value of its expression argument.
4444 - Now bdist_rpm distutils command understands a verify_script option in
4445 the config file, including the contents of the referred filename in
4446 the "%verifyscript" section of the rpm spec file.
4448 - Fixed bug #495695: webbrowser module would run graphic browsers in a
4449 unix environment even if DISPLAY was not set. Also, support for
4450 skipstone browser was included.
4452 - Fixed bug #636769: rexec would run unallowed code if subclasses of
4453 strings were used as parameters for certain functions.
4458 - pygettext.py now supports globbing on Windows, and accepts module
4459 names in addition to accepting file names.
4461 - The SGI demos (Demo/sgi) have been removed. Nobody thought they
4462 were interesting any more. (The SGI library modules and extensions
4463 are still there; it is believed that at least some of these are
4464 still used and useful.)
4466 - IDLE supports the new encoding declarations (PEP 263); it can also
4467 deal with legacy 8-bit files if they use the locale's encoding. It
4468 allows non-ASCII strings in the interactive shell and executes them
4469 in the locale's encoding.
4471 - freeze.py now produces binaries which can import shared modules,
4472 unlike before when this failed due to missing symbol exports in
4473 the generated binary.
4478 - On Unix, IDLE is now installed automatically.
4480 - The fpectl module is not built by default; it's dangerous or useless
4481 except in the hands of experts.
4483 - The public Python C API will generally be declared using PyAPI_FUNC
4484 and PyAPI_DATA macros, while Python extension module init functions
4485 will be declared with PyMODINIT_FUNC. DL_EXPORT/DL_IMPORT macros
4488 - A bug was fixed that could cause COUNT_ALLOCS builds to segfault, or
4489 get into infinite loops, when a new-style class got garbage-collected.
4490 Unfortunately, to avoid this, the way COUNT_ALLOCS works requires
4491 that new-style classes be immortal in COUNT_ALLOCS builds. Note that
4492 COUNT_ALLOCS is not enabled by default, in either release or debug
4493 builds, and that new-style classes are immortal only in COUNT_ALLOCS
4496 - Compiling out the cyclic garbage collector is no longer an option.
4497 The old symbol WITH_CYCLE_GC is now ignored, and Python.h arranges
4498 that it's always defined (for the benefit of any extension modules
4499 that may be conditionalizing on it). A bonus is that any extension
4500 type participating in cyclic gc can choose to participate in the
4501 Py_TRASHCAN mechanism now too; in the absence of cyclic gc, this used
4502 to require editing the core to teach the trashcan mechanism about the
4505 - According to Annex F of the current C standard,
4507 The Standard C macro HUGE_VAL and its float and long double analogs,
4508 HUGE_VALF and HUGE_VALL, expand to expressions whose values are
4509 positive infinities.
4511 Python only uses the double HUGE_VAL, and only to #define its own symbol
4512 Py_HUGE_VAL. Some platforms have incorrect definitions for HUGE_VAL.
4513 pyport.h used to try to worm around that, but the workarounds triggered
4514 other bugs on other platforms, so we gave up. If your platform defines
4515 HUGE_VAL incorrectly, you'll need to #define Py_HUGE_VAL to something
4516 that works on your platform. The only instance of this I'm sure about
4517 is on an unknown subset of Cray systems, described here:
4519 http://www.cray.com/swpubs/manuals/SN-2194_2.0/html-SN-2194_2.0/x3138.htm
4521 Presumably 2.3a1 breaks such systems. If anyone uses such a system, help!
4523 - The configure option --without-doc-strings can be used to remove the
4524 doc strings from the builtin functions and modules; this reduces the
4525 size of the executable.
4527 - The universal newlines option (PEP 278) is on by default. On Unix
4528 it can be disabled by passing --without-universal-newlines to the
4529 configure script. On other platforms, remove
4530 WITH_UNIVERSAL_NEWLINES from pyconfig.h.
4532 - On Unix, a shared libpython2.3.so can be created with --enable-shared.
4534 - All uses of the CACHE_HASH, INTERN_STRINGS, and DONT_SHARE_SHORT_STRINGS
4535 preprocessor symbols were eliminated. The internal decisions they
4536 controlled stopped being experimental long ago.
4538 - The tools used to build the documentation now work under Cygwin as
4541 - The bsddb and dbm module builds have been changed to try and avoid version
4542 skew problems and disable linkage with Berkeley DB 1.85 unless the
4543 installer knows what s/he's doing. See the section on building these
4544 modules in the README file for details.
4549 - PyNumber_Check() now returns true for string and unicode objects.
4550 This is a result of these types having a partially defined
4551 tp_as_number slot. (This is not a feature, but an indication that
4552 PyNumber_Check() is not very useful to determine numeric behavior.
4553 It may be deprecated.)
4555 - The string object's layout has changed: the pointer member
4556 ob_sinterned has been replaced by an int member ob_sstate. On some
4557 platforms (e.g. most 64-bit systems) this may change the offset of
4558 the ob_sval member, so as a precaution the API_VERSION has been
4559 incremented. The apparently unused feature of "indirect interned
4560 strings", supported by the ob_sinterned member, is gone. Interned
4561 strings are now usually mortal; there is a new API,
4562 PyString_InternImmortal() that creates immortal interned strings.
4563 (The ob_sstate member can only take three values; however, while
4564 making it a char saves a few bytes per string object on average, in
4565 it also slowed things down a bit because ob_sval was no longer
4568 - The Py_InitModule*() functions now accept NULL for the 'methods'
4569 argument. Modules without global functions are becoming more common
4570 now that factories can be types rather than functions.
4572 - New C API PyUnicode_FromOrdinal() which exposes unichr() at C
4575 - New functions PyErr_SetExcFromWindowsErr() and
4576 PyErr_SetExcFromWindowsErrWithFilename(). Similar to
4577 PyErr_SetFromWindowsErrWithFilename() and
4578 PyErr_SetFromWindowsErr(), but they allow to specify
4579 the exception type to raise. Available on Windows.
4581 - Py_FatalError() is now declared as taking a const char* argument. It
4582 was previously declared without const. This should not affect working
4585 - Added new macro PySequence_ITEM(o, i) that directly calls
4586 sq_item without rechecking that o is a sequence and without
4587 adjusting for negative indices.
4589 - PyRange_New() now raises ValueError if the fourth argument is not 1.
4590 This is part of the removal of deprecated features of the xrange
4593 - PyNumber_Coerce() and PyNumber_CoerceEx() now also invoke the type's
4594 coercion if both arguments have the same type but this type has the
4595 CHECKTYPES flag set. This is to better support proxies.
4597 - The type of tp_free has been changed from "``void (*)(PyObject *)``" to
4598 "``void (*)(void *)``".
4600 - PyObject_Del, PyObject_GC_Del are now functions instead of macros.
4602 - A type can now inherit its metatype from its base type. Previously,
4603 when PyType_Ready() was called, if ob_type was found to be NULL, it
4604 was always set to &PyType_Type; now it is set to base->ob_type,
4605 where base is tp_base, defaulting to &PyObject_Type.
4607 - PyType_Ready() accidentally did not inherit tp_is_gc; now it does.
4609 - The PyCore_* family of APIs have been removed.
4611 - The "u#" parser marker will now pass through Unicode objects as-is
4612 without going through the buffer API.
4614 - The enumerators of cmp_op have been renamed to use the prefix ``PyCmp_``.
4616 - An old #define of ANY as void has been removed from pyport.h. This
4617 hasn't been used since Python's pre-ANSI days, and the #define has
4618 been marked as obsolete since then. SF bug 495548 says it created
4619 conflicts with other packages, so keeping it around wasn't harmless.
4621 - Because Python's magic number scheme broke on January 1st, we decided
4622 to stop Python development. Thanks for all the fish!
4624 - Some of us don't like fish, so we changed Python's magic number
4625 scheme to a new one. See Python/import.c for details.
4630 - OpenVMS is now supported.
4632 - AtheOS is now supported.
4634 - the EMX runtime environment on OS/2 is now supported.
4636 - GNU/Hurd is now supported.
4641 - The regrtest.py script's -u option now provides a way to say "allow
4642 all resources except this one." For example, to allow everything
4643 except bsddb, give the option '-uall,-bsddb'.
4648 - The Windows distribution now ships with version 4.0.14 of the
4649 Sleepycat Berkeley database library. This should be a huge
4650 improvement over the previous Berkeley DB 1.85, which had many
4652 XXX What are the licensing issues here?
4653 XXX If a user has a database created with a previous version of
4654 XXX Python, what must they do to convert it?
4655 XXX I'm still not sure how to link this thing (see PCbuild/readme.txt).
4656 XXX The version # is likely to change before 2.3a1.
4658 - The Windows distribution now ships with a Secure Sockets Library (SLL)
4661 - The Windows distribution now ships with Tcl/Tk version 8.4.1 (it
4662 previously shipped with Tcl/Tk 8.3.2).
4664 - When Python is built under a Microsoft compiler, sys.version now
4665 includes the compiler version number (_MSC_VER). For example, under
4666 MSVC 6, sys.version contains the substring "MSC v.1200 ". 1200 is
4667 the value of _MSC_VER under MSVC 6.
4669 - Sometimes the uninstall executable (UNWISE.EXE) vanishes. One cause
4670 of that has been fixed in the installer (disabled Wise's "delete in-
4671 use files" uninstall option).
4673 - Fixed a bug in urllib's proxy handling in Windows. [SF bug #503031]
4675 - The installer now installs Start menu shortcuts under (the local
4676 equivalent of) "All Users" when doing an Admin install.
4678 - file.truncate([newsize]) now works on Windows for all newsize values.
4679 It used to fail if newsize didn't fit in 32 bits, reflecting a
4680 limitation of MS _chsize (which is no longer used).
4682 - os.waitpid() is now implemented for Windows, and can be used to block
4683 until a specified process exits. This is similar to, but not exactly
4684 the same as, os.waitpid() on POSIX systems. If you're waiting for
4685 a specific process whose pid was obtained from one of the spawn()
4686 functions, the same Python os.waitpid() code works across platforms.
4687 See the docs for details. The docs were changed to clarify that
4688 spawn functions return, and waitpid requires, a process handle on
4689 Windows (not the same thing as a Windows process id).
4691 - New tempfile.TemporaryFile implementation for Windows: this doesn't
4692 need a TemporaryFileWrapper wrapper anymore, and should be immune
4693 to a nasty problem: before 2.3, if you got a temp file on Windows, it
4694 got wrapped in an object whose close() method first closed the
4695 underlying file, then deleted the file. This usually worked fine.
4696 However, the spawn family of functions on Windows create (at a low C
4697 level) the same set of open files in the spawned process Q as were
4698 open in the spawning process P. If a temp file f was among them, then
4699 doing f.close() in P first closed P's C-level file handle on f, but Q's
4700 C-level file handle on f remained open, so the attempt in P to delete f
4701 blew up with a "Permission denied" error (Windows doesn't allow
4702 deleting open files). This was surprising, subtle, and difficult to
4705 - The os module now exports all the symbolic constants usable with the
4706 low-level os.open() on Windows: the new constants in 2.3 are
4707 O_NOINHERIT, O_SHORT_LIVED, O_TEMPORARY, O_RANDOM and O_SEQUENTIAL.
4708 The others were also available in 2.2: O_APPEND, O_BINARY, O_CREAT,
4709 O_EXCL, O_RDONLY, O_RDWR, O_TEXT, O_TRUNC and O_WRONLY. Contrary
4710 to Microsoft docs, O_SHORT_LIVED does not seem to imply O_TEMPORARY
4711 (so specify both if you want both; note that neither is useful unless
4712 specified with O_CREAT too).
4717 - Mac/Relnotes is gone, the release notes are now here.
4719 - Python (the OSX-only, unix-based version, not the OS9-compatible CFM
4720 version) now fully supports unicode strings as arguments to various file
4721 system calls, eg. open(), file(), os.stat() and os.listdir().
4723 - The current naming convention for Python on the Macintosh is that MacPython
4724 refers to the unix-based OSX-only version, and MacPython-OS9 refers to the
4725 CFM-based version that runs on both OS9 and OSX.
4727 - All MacPython-OS9 functionality is now available in an OSX unix build,
4728 including the Carbon modules, the IDE, OSA support, etc. A lot of this
4729 will only work correctly in a framework build, though, because you cannot
4730 talk to the window manager unless your application is run from a .app
4731 bundle. There is a command line tool "pythonw" that runs your script
4732 with an interpreter living in such a .app bundle, this interpreter should
4733 be used to run any Python script using the window manager (including
4734 Tkinter or wxPython scripts).
4736 - Most of Mac/Lib has moved to Lib/plat-mac, which is again used both in
4737 MacPython-OSX and MacPython-OS9. The only modules remaining in Mac/Lib
4738 are specifically for MacPython-OS9 (CFM support, preference resources, etc).
4740 - A new utility PythonLauncher will start a Python interpreter when a .py or
4741 .pyw script is double-clicked in the Finder. By default .py scripts are
4742 run with a normal Python interpreter in a Terminal window and .pyw
4743 files are run with a window-aware pythonw interpreter without a Terminal
4744 window, but all this can be customized.
4746 - MacPython-OS9 is now Carbon-only, so it runs on Mac OS 9 or Mac OS X and
4747 possibly on Mac OS 8.6 with the right CarbonLib installed, but not on earlier
4750 - Many tools such as BuildApplet.py and gensuitemodule.py now support a command
4753 - All the Carbon classes are now PEP253 compliant, meaning that you can
4754 subclass them from Python. Most of the attributes have gone, you should
4755 now use the accessor function call API, which is also what Apple's
4756 documentation uses. Some attributes such as grafport.visRgn are still
4757 available for convenience.
4759 - New Carbon modules File (implementing the APIs in Files.h and Aliases.h)
4760 and Folder (APIs from Folders.h). The old macfs builtin module is
4761 gone, and replaced by a Python wrapper around the new modules.
4763 - Pathname handling should now be fully consistent: MacPython-OSX always uses
4764 unix pathnames and MacPython-OS9 always uses colon-separated Mac pathnames
4765 (also when running on Mac OS X).
4767 - New Carbon modules Help and AH give access to the Carbon Help Manager.
4768 There are hooks in the IDE to allow accessing the Python documentation
4769 (and Apple's Carbon and Cocoa documentation) through the Help Viewer.
4770 See Mac/OSX/README for converting the Python documentation to a
4771 Help Viewer compatible form and installing it.
4773 - OSA support has been redesigned and the generated Python classes now
4774 mirror the inheritance defined by the underlying OSA classes.
4776 - MacPython no longer maps both \r and \n to \n on input for any text file.
4777 This feature has been replaced by universal newline support (PEP278).
4779 - The default encoding for Python sourcefiles in MacPython-OS9 is no longer
4780 mac-roman (or whatever your local Mac encoding was) but "ascii", like on
4781 other platforms. If you really need sourcefiles with Mac characters in them
4782 you can change this in site.py.
4785 What's New in Python 2.2 final?
4786 ===============================
4788 *Release date: 21-Dec-2001*
4790 Type/class unification and new-style classes
4791 --------------------------------------------
4793 - pickle.py, cPickle: allow pickling instances of new-style classes
4794 with a custom metaclass.
4799 - weakref proxy object: when comparing, unwrap both arguments if both
4805 - binascii.b2a_base64(): fix a potential buffer overrun when encoding
4808 - cPickle: the obscure "fast" mode was suspected of causing stack
4809 overflows on the Mac. Hopefully fixed this by setting the recursion
4810 limit much smaller. If the limit is too low (it only affects
4811 performance), you can change it by defining PY_CPICKLE_FAST_LIMIT
4812 when compiling cPickle.c (or in pyconfig.h).
4817 - dumbdbm.py: fixed a dumb old bug (the file didn't get synched at
4818 close or delete time).
4820 - rfc822.py: fixed a bug where the address '<>' was converted to None
4821 instead of an empty string (also fixes the email.Utils module).
4823 - xmlrpclib.py: version 1.0.0; uses precision for doubles.
4825 - test suite: the pickle and cPickle tests were not executing any code
4826 when run from the standard regression test.
4846 - distutils package: fixed broken Windows installers (bdist_wininst).
4848 - tempfile.py: prevent mysterious warnings when TemporaryFileWrapper
4849 instances are deleted at process exit time.
4851 - socket.py: prevent mysterious warnings when socket instances are
4852 deleted at process exit time.
4854 - posixmodule.c: fix a Windows crash with stat() of a filename ending
4860 - The Carbon toolbox modules have been upgraded to Universal Headers
4861 3.4, and experimental CoreGraphics and CarbonEvents modules have
4862 been added. All only for framework-enabled MacOSX.
4865 What's New in Python 2.2c1?
4866 ===========================
4868 *Release date: 14-Dec-2001*
4870 Type/class unification and new-style classes
4871 --------------------------------------------
4873 - Guido's tutorial introduction to the new type/class features has
4874 been extensively updated. See
4876 http://www.python.org/2.2/descrintro.html
4878 That remains the primary documentation in this area.
4880 - Fixed a leak: instance variables declared with __slots__ were never
4883 - The "delete attribute" method of descriptor objects is called
4884 __delete__, not __del__. In previous releases, it was mistakenly
4885 called __del__, which created an unfortunate overloading condition
4886 with finalizers. (The "get attribute" and "set attribute" methods
4887 are still called __get__ and __set__, respectively.)
4889 - Some subtle issues with the super built-in were fixed:
4891 (a) When super itself is subclassed, its __get__ method would still
4892 return an instance of the base class (i.e., of super).
4894 (b) super(C, C()).__class__ would return C rather than super. This
4895 is confusing. To fix this, I decided to change the semantics of
4896 super so that it only applies to code attributes, not to data
4897 attributes. After all, overriding data attributes is not
4900 (c) The __get__ method didn't check whether the argument was an
4901 instance of the type used in creation of the super instance.
4903 - Previously, hash() of an instance of a subclass of a mutable type
4904 (list or dictionary) would return some value, rather than raising
4905 TypeError. This has been fixed. Also, directly calling
4906 dict.__hash__ and list.__hash__ now raises the same TypeError
4907 (previously, these were the same as object.__hash__).
4909 - New-style objects now support deleting their __dict__. This is for
4910 all intents and purposes equivalent to assigning a brand new empty
4911 dictionary, but saves space if the object is not used further.
4916 - -Qnew now works as documented in PEP 238: when -Qnew is passed on
4917 the command line, all occurrences of "/" use true division instead
4918 of classic division. See the PEP for details. Note that "all"
4919 means all instances in library and 3rd-party modules, as well as in
4920 your own code. As the PEP says, -Qnew is intended for use only in
4921 educational environments with control over the libraries in use.
4922 Note that test_coercion.py in the standard Python test suite fails
4923 under -Qnew; this is expected, and won't be repaired until true
4924 division becomes the default (in the meantime, test_coercion is
4925 testing the current rules).
4927 - complex() now only allows the first argument to be a string
4928 argument, and raises TypeError if either the second arg is a string
4929 or if the second arg is specified when the first is a string.
4934 - gc.get_referents was renamed to gc.get_referrers.
4939 - Functions in the os.spawn() family now release the global interpreter
4940 lock around calling the platform spawn. They should always have done
4941 this, but did not before 2.2c1. Multithreaded programs calling
4942 an os.spawn function with P_WAIT will no longer block all Python threads
4943 until the spawned program completes. It's possible that some programs
4944 relies on blocking, although more likely by accident than by design.
4946 - webbrowser defaults to netscape.exe on OS/2 now.
4948 - Tix.ResizeHandle exposes detach_widget, hide, and show.
4950 - The charset alias windows_1252 has been added.
4952 - types.StringTypes is a tuple containing the defined string types;
4953 usually this will be (str, unicode), but if Python was compiled
4954 without Unicode support it will be just (str,).
4956 - The pulldom and minidom modules were synchronized to PyXML.
4961 - A new script called Tools/scripts/google.py was added, which fires
4962 off a search on Google.
4967 - Note that release builds of Python should arrange to define the
4968 preprocessor symbol NDEBUG on the command line (or equivalent).
4969 In the 2.2 pre-release series we tried to define this by magic in
4970 Python.h instead, but it proved to cause problems for extension
4971 authors. The Unix, Windows and Mac builds now all define NDEBUG in
4972 release builds via cmdline (or equivalent) instead. Ports to
4973 other platforms should do likewise.
4975 - It is no longer necessary to use --with-suffix when building on a
4976 case-insensitive file system (such as Mac OS X HFS+). In the build
4977 directory an extension is used, but not in the installed python.
4982 - New function PyDict_MergeFromSeq2() exposes the builtin dict
4983 constructor's logic for updating a dictionary from an iterable object
4984 producing key-value pairs.
4986 - PyArg_ParseTupleAndKeywords() requires that the number of entries in
4987 the keyword list equal the number of argument specifiers. This
4988 wasn't checked correctly, and PyArg_ParseTupleAndKeywords could even
4989 dump core in some bad cases. This has been repaired. As a result,
4990 PyArg_ParseTupleAndKeywords may raise RuntimeError in bad cases that
4991 previously went unchallenged.
5005 - In unix-Python on Mac OS X (and darwin) sys.platform is now "darwin",
5006 without any trailing digits.
5008 - Changed logic for finding python home in Mac OS X framework Pythons.
5009 Now sys.executable points to the executable again, in stead of to
5010 the shared library. The latter is used only for locating the python
5014 What's New in Python 2.2b2?
5015 ===========================
5017 *Release date: 16-Nov-2001*
5019 Type/class unification and new-style classes
5020 --------------------------------------------
5022 - Multiple inheritance mixing new-style and classic classes in the
5023 list of base classes is now allowed, so this works now:
5026 class Mixed(Classic, object): pass
5028 The MRO (method resolution order) for each base class is respected
5029 according to its kind, but the MRO for the derived class is computed
5030 using new-style MRO rules if any base class is a new-style class.
5031 This needs to be documented.
5033 - The new builtin dictionary() constructor, and dictionary type, have
5034 been renamed to dict. This reflects a decade of common usage.
5036 - dict() now accepts an iterable object producing 2-sequences. For
5037 example, dict(d.items()) == d for any dictionary d. The argument,
5038 and the elements of the argument, can be any iterable objects.
5040 - New-style classes can now have a __del__ method, which is called
5041 when the instance is deleted (just like for classic classes).
5043 - Assignment to object.__dict__ is now possible, for objects that are
5044 instances of new-style classes that have a __dict__ (unless the base
5047 - Methods of built-in types now properly check for keyword arguments
5048 (formerly these were silently ignored). The only built-in methods
5049 that take keyword arguments are __call__, __init__ and __new__.
5051 - The socket function has been converted to a type; see below.
5056 - Assignment to __debug__ raises SyntaxError at compile-time. This
5057 was promised when 2.1c1 was released as "What's New in Python 2.1c1"
5060 - Clarified the error messages for unsupported operands to an operator
5066 - mmap has a new keyword argument, "access", allowing a uniform way for
5067 both Windows and Unix users to create read-only, write-through and
5068 copy-on-write memory mappings. This was previously possible only on
5069 Unix. A new keyword argument was required to support this in a
5070 uniform way because the mmap() signatures had diverged across
5071 platforms. Thanks to Jay T Miller for repairing this!
5073 - By default, the gc.garbage list now contains only those instances in
5074 unreachable cycles that have __del__ methods; in 2.1 it contained all
5075 instances in unreachable cycles. "Instances" here has been generalized
5076 to include instances of both new-style and old-style classes.
5078 - The socket module defines a new method for socket objects,
5079 sendall(). This is like send() but may make multiple calls to
5080 send() until all data has been sent. Also, the socket function has
5081 been converted to a subclassable type, like list and tuple (etc.)
5082 before it; socket and SocketType are now the same thing.
5084 - Various bugfixes to the curses module. There is now a test suite
5085 for the curses module (you have to run it manually).
5087 - binascii.b2a_base64 no longer places an arbitrary restriction of 57
5093 - tkFileDialog exposes a Directory class and askdirectory
5094 convenience function.
5096 - Symbolic group names in regular expressions must be unique. For
5097 example, the regexp r'(?P<abc>)(?P<abc>)' is not allowed, because a
5098 single name can't mean both "group 1" and "group 2" simultaneously.
5099 Python 2.2 detects this error at regexp compilation time;
5100 previously, the error went undetected, and results were
5101 unpredictable. Also in sre, the pattern.split(), pattern.sub(), and
5102 pattern.subn() methods have been rewritten in C. Also, an
5103 experimental function/method finditer() has been added, which works
5104 like findall() but returns an iterator.
5106 - Tix exposes more commands through the classes DirSelectBox,
5107 DirSelectDialog, ListNoteBook, Meter, CheckList, and the
5108 methods tix_addbitmapdir, tix_cget, tix_configure, tix_filedialog,
5109 tix_getbitmap, tix_getimage, tix_option_get, and tix_resetoptions.
5111 - Traceback objects are now scanned by cyclic garbage collection, so
5112 cycles created by casual use of sys.exc_info() no longer cause
5113 permanent memory leaks (provided garbage collection is enabled).
5115 - os.extsep -- a new variable needed by the RISCOS support. It is the
5116 separator used by extensions, and is '.' on all platforms except
5117 RISCOS, where it is '/'. There is no need to use this variable
5118 unless you have a masochistic desire to port your code to RISCOS.
5120 - mimetypes.py has optional support for non-standard, but commonly
5121 found types. guess_type() and guess_extension() now accept an
5122 optional 'strict' flag, defaulting to true, which controls whether
5123 recognize non-standard types or not. A few non-standard types we
5124 know about have been added. Also, when run as a script, there are
5125 new -l and -e options.
5127 - statcache is now deprecated.
5129 - email.Utils.formatdate() now produces the preferred RFC 2822 style
5130 dates with numeric timezones (it used to produce obsolete dates
5131 hard coded to "GMT" timezone). An optional 'localtime' flag is
5132 added to produce dates in the local timezone, with daylight savings
5133 time properly taken into account.
5135 - In pickle and cPickle, instead of masking errors in load() by
5136 transforming them into SystemError, we let the original exception
5137 propagate out. Also, implement support for __safe_for_unpickling__
5138 in pickle, as it already was supported in cPickle.
5146 - The dbm module is built using libdb1 if available. The bsddb module
5147 is built with libdb3 if available.
5149 - Misc/Makefile.pre.in has been removed by BDFL pronouncement.
5154 - New function PySequence_Fast_GET_SIZE() returns the size of a non-
5155 NULL result from PySequence_Fast(), more quickly than calling
5158 - New argument unpacking function PyArg_UnpackTuple() added.
5160 - New functions PyObject_CallFunctionObjArgs() and
5161 PyObject_CallMethodObjArgs() have been added to make it more
5162 convenient and efficient to call functions and methods from C.
5164 - PyArg_ParseTupleAndKeywords() no longer masks errors, so it's
5165 possible that this will propagate errors it didn't before.
5167 - New function PyObject_CheckReadBuffer(), which returns true if its
5168 argument supports the single-segment readable buffer interface.
5173 - We've finally confirmed that this release builds on HP-UX 11.00,
5174 *with* threads, and passes the test suite.
5176 - Thanks to a series of patches from Michael Muller, Python may build
5177 again under OS/2 Visual Age C++.
5179 - Updated RISCOS port by Dietmar Schwertberger.
5184 - Added a test script for the curses module. It isn't run automatically;
5185 regrtest.py must be run with '-u curses' to enable it.
5193 - PythonScript has been moved to unsupported and is slated to be
5194 removed completely in the next release.
5196 - It should now be possible to build applets that work on both OS9 and
5199 - The core is now linked with CoreServices not Carbon; as a side
5200 result, default 8bit encoding on OSX is now ASCII.
5202 - Python should now build on OSX 10.1.1
5205 What's New in Python 2.2b1?
5206 ===========================
5208 *Release date: 19-Oct-2001*
5210 Type/class unification and new-style classes
5211 --------------------------------------------
5213 - New-style classes are now always dynamic (except for built-in and
5214 extension types). There is no longer a performance penalty, and I
5215 no longer see another reason to keep this baggage around. One relic
5216 remains: the __dict__ of a new-style class is a read-only proxy; you
5217 must set the class's attribute to modify it. As a consequence, the
5218 __defined__ attribute of new-style types no longer exists, for lack
5219 of need: there is once again only one __dict__ (although in the
5220 future a __cache__ may be resurrected with a similar function, if I
5221 can prove that it actually speeds things up).
5223 - C.__doc__ now works as expected for new-style classes (in 2.2a4 it
5224 always returned None, even when there was a class docstring).
5226 - doctest now finds and runs docstrings attached to new-style classes,
5227 class methods, static methods, and properties.
5232 - A very subtle syntactical pitfall in list comprehensions was fixed.
5233 For example: [a+b for a in 'abc', for b in 'def']. The comma in
5234 this example is a mistake. Previously, this would silently let 'a'
5235 iterate over the singleton tuple ('abc',), yielding ['abcd', 'abce',
5236 'abcf'] rather than the intended ['ad', 'ae', 'af', 'bd', 'be',
5237 'bf', 'cd', 'ce', 'cf']. Now, this is flagged as a syntax error.
5238 Note that [a for a in <singleton>] is a convoluted way to say
5239 [<singleton>] anyway, so it's not like any expressiveness is lost.
5241 - getattr(obj, name, default) now only catches AttributeError, as
5242 documented, rather than returning the default value for all
5243 exceptions (which could mask bugs in a __getattr__ hook, for
5246 - Weak reference objects are now part of the core and offer a C API.
5247 A bug which could allow a core dump when binary operations involved
5248 proxy reference has been fixed. weakref.ReferenceError is now a
5251 - unicode(obj) now behaves more like str(obj), accepting arbitrary
5252 objects, and calling a __unicode__ method if it exists.
5253 unicode(obj, encoding) and unicode(obj, encoding, errors) still
5254 require an 8-bit string or character buffer argument.
5256 - isinstance() now allows any object as the first argument and a
5257 class, a type or something with a __bases__ tuple attribute for the
5258 second argument. The second argument may also be a tuple of a
5259 class, type, or something with __bases__, in which case isinstance()
5260 will return true if the first argument is an instance of any of the
5261 things contained in the second argument tuple. E.g.
5263 isinstance(x, (A, B))
5265 returns true if x is an instance of A or B.
5270 - thread.start_new_thread() now returns the thread ID (previously None).
5272 - binascii has now two quopri support functions, a2b_qp and b2a_qp.
5274 - readline now supports setting the startup_hook and the
5275 pre_event_hook, and adds the add_history() function.
5277 - os and posix supports chroot(), setgroups() and unsetenv() where
5278 available. The stat(), fstat(), statvfs() and fstatvfs() functions
5279 now return "pseudo-sequences" -- the various fields can now be
5280 accessed as attributes (e.g. os.stat("/").st_mtime) but for
5281 backwards compatibility they also behave as a fixed-length sequence.
5282 Some platform-specific fields (e.g. st_rdev) are only accessible as
5285 - time: localtime(), gmtime() and strptime() now return a
5286 pseudo-sequence similar to the os.stat() return value, with
5287 attributes like tm_year etc.
5289 - Decompression objects in the zlib module now accept an optional
5290 second parameter to decompress() that specifies the maximum amount
5291 of memory to use for the uncompressed data.
5293 - optional SSL support in the socket module now exports OpenSSL
5294 functions RAND_add(), RAND_egd(), and RAND_status(). These calls
5295 are useful on platforms like Solaris where OpenSSL does not
5296 automatically seed its PRNG. Also, the keyfile and certfile
5297 arguments to socket.ssl() are now optional.
5299 - posixmodule (and by extension, the os module on POSIX platforms) now
5300 exports O_LARGEFILE, O_DIRECT, O_DIRECTORY, and O_NOFOLLOW.
5305 - doctest now excludes functions and classes not defined by the module
5306 being tested, thanks to Tim Hochberg.
5308 - HotShot, a new profiler implemented using a C-based callback, has
5309 been added. This substantially reduces the overhead of profiling,
5310 but it is still quite preliminary. Support modules and
5311 documentation will be added in upcoming releases (before 2.2 final).
5313 - profile now produces correct output in situations where an exception
5314 raised in Python is cleared by C code (e.g. hasattr()). This used
5315 to cause wrong output, including spurious claims of recursive
5316 functions and attribution of time spent to the wrong function.
5318 The code and documentation for the derived OldProfile and HotProfile
5319 profiling classes was removed. The code hasn't worked for years (if
5320 you tried to use them, they raised exceptions). OldProfile
5321 intended to reproduce the behavior of the profiler Python used more
5322 than 7 years ago, and isn't interesting anymore. HotProfile intended
5323 to provide a faster profiler (but producing less information), and
5324 that's a worthy goal we intend to meet via a different approach (but
5325 without losing information).
5327 - Profile.calibrate() has a new implementation that should deliver
5328 a much better system-specific calibration constant. The constant can
5329 now be specified in an instance constructor, or as a Profile class or
5330 instance variable, instead of by editing profile.py's source code.
5331 Calibration must still be done manually (see the docs for the profile
5334 Note that Profile.calibrate() must be overridden by subclasses.
5335 Improving the accuracy required exploiting detailed knowledge of
5336 profiler internals; the earlier method abstracted away the details
5337 and measured a simplified model instead, but consequently computed
5338 a constant too small by a factor of 2 on some modern machines.
5340 - quopri's encode and decode methods take an optional header parameter,
5341 which indicates whether output is intended for the header 'Q'
5344 - The SocketServer.ThreadingMixIn class now closes the request after
5345 finish_request() returns. (Not when it errors out though.)
5347 - The nntplib module's NNTP.body() method has grown a 'file' argument
5348 to allow saving the message body to a file.
5350 - The email package has added a class email.Parser.HeaderParser which
5351 only parses headers and does not recurse into the message's body.
5352 Also, the module/class MIMEAudio has been added for representing
5353 audio data (contributed by Anthony Baxter).
5355 - ftplib should be able to handle files > 2GB.
5357 - ConfigParser.getboolean() now also interprets TRUE, FALSE, YES, NO,
5360 - xml.dom.minidom NodeList objects now support the length attribute
5361 and item() method as required by the DOM specifications.
5366 - Demo/dns was removed. It no longer serves any purpose; a package
5367 derived from it is now maintained by Anthony Baxter, see
5368 http://PyDNS.SourceForge.net.
5370 - The freeze tool has been made more robust, and two new options have
5371 been added: -X and -E.
5376 - configure will use CXX in LINKCC if CXX is used to build main() and
5377 the system requires to link a C++ main using the C++ compiler.
5382 - The documentation for the tp_compare slot is updated to require that
5383 the return value must be -1, 0, 1; an arbitrary number <0 or >0 is
5384 not correct. This is not yet enforced but will be enforced in
5385 Python 2.3; even later, we may use -2 to indicate errors and +2 for
5386 "NotImplemented". Right now, -1 should be used for an error return.
5388 - PyLong_AsLongLong() now accepts int (as well as long) arguments.
5389 Consequently, PyArg_ParseTuple's 'L' code also accepts int (as well
5392 - PyThread_start_new_thread() now returns a long int giving the thread
5393 ID, if one can be calculated; it returns -1 for error, 0 if no
5394 thread ID is calculated (this is an incompatible change, but only
5395 the thread module used this API). This code has only really been
5396 tested on Linux and Windows; other platforms please beware (and
5397 report any bugs or strange behavior).
5399 - PyUnicode_FromEncodedObject() no longer accepts Unicode objects as
5411 - Installer: If you install IDLE, and don't disable file-extension
5412 registration, a new "Edit with IDLE" context (right-click) menu entry
5413 is created for .py and .pyw files.
5415 - The signal module now supports SIGBREAK on Windows, thanks to Steven
5416 Scott. Note that SIGBREAK is unique to Windows. The default SIGBREAK
5417 action remains to call Win32 ExitProcess(). This can be changed via
5418 signal.signal(). For example::
5420 # Make Ctrl+Break raise KeyboardInterrupt, like Python's default Ctrl+C
5421 # (SIGINT) behavior.
5423 signal.signal(signal.SIGBREAK, signal.default_int_handler)
5428 except KeyboardInterrupt:
5429 # We get here on Ctrl+C or Ctrl+Break now; if we had not changed
5430 # SIGBREAK, only on Ctrl+C (and Ctrl+Break would terminate the
5431 # program without the possibility for any Python-level cleanup).
5435 What's New in Python 2.2a4?
5436 ===========================
5438 *Release date: 28-Sep-2001*
5440 Type/class unification and new-style classes
5441 --------------------------------------------
5443 - pydoc and inspect are now aware of new-style classes;
5444 e.g. help(list) at the interactive prompt now shows proper
5445 documentation for all operations on list objects.
5447 - Applications using Jim Fulton's ExtensionClass module can now safely
5448 be used with Python 2.2. In particular, Zope 2.4.1 now works with
5449 Python 2.2 (as well as with Python 2.1.1). The Demo/metaclass
5450 examples also work again. It is hoped that Gtk and Boost also work
5451 with 2.2a4 and beyond. (If you can confirm this, please write
5452 webmaster@python.org; if there are still problems, please open a bug
5453 report on SourceForge.)
5455 - property() now takes 4 keyword arguments: fget, fset, fdel and doc.
5456 These map to read-only attributes 'fget', 'fset', 'fdel', and '__doc__'
5457 in the constructed property object. fget, fset and fdel weren't
5458 discoverable from Python in 2.2a3. __doc__ is new, and allows to
5459 associate a docstring with a property.
5461 - Comparison overloading is now more completely implemented. For
5462 example, a str subclass instance can properly be compared to a str
5463 instance, and it can properly overload comparison. Ditto for most
5464 other built-in object types.
5466 - The repr() of new-style classes has changed; instead of <type
5467 'M.Foo'> a new-style class is now rendered as <class 'M.Foo'>,
5468 *except* for built-in types, which are still rendered as <type
5469 'Foo'> (to avoid upsetting existing code that might parse or
5470 otherwise rely on repr() of certain type objects).
5472 - The repr() of new-style objects is now always <Foo object at XXX>;
5473 previously, it was sometimes <Foo instance at XXX>.
5475 - For new-style classes, what was previously called __getattr__ is now
5476 called __getattribute__. This method, if defined, is called for
5477 *every* attribute access. A new __getattr__ hook more similar to the
5478 one in classic classes is defined which is called only if regular
5479 attribute access raises AttributeError; to catch *all* attribute
5480 access, you can use __getattribute__ (for new-style classes). If
5481 both are defined, __getattribute__ is called first, and if it raises
5482 AttributeError, __getattr__ is called.
5484 - The __class__ attribute of new-style objects can be assigned to.
5485 The new class must have the same C-level object layout as the old
5488 - The builtin file type can be subclassed now. In the usual pattern,
5489 "file" is the name of the builtin type, and file() is a new builtin
5490 constructor, with the same signature as the builtin open() function.
5491 file() is now the preferred way to open a file.
5493 - Previously, __new__ would only see sequential arguments passed to
5494 the type in a constructor call; __init__ would see both sequential
5495 and keyword arguments. This made no sense whatsoever any more, so
5496 now both __new__ and __init__ see all arguments.
5498 - Previously, hash() applied to an instance of a subclass of str or
5499 unicode always returned 0. This has been repaired.
5501 - Previously, an operation on an instance of a subclass of an
5502 immutable type (int, long, float, complex, tuple, str, unicode),
5503 where the subtype didn't override the operation (and so the
5504 operation was handled by the builtin type), could return that
5505 instance instead a value of the base type. For example, if s was of
5506 a str subclass type, s[:] returned s as-is. Now it returns a str
5507 with the same value as s.
5509 - Provisional support for pickling new-style objects has been added.
5514 - file.writelines() now accepts any iterable object producing strings.
5516 - PyUnicode_FromEncodedObject() now works very much like
5517 PyObject_Str(obj) in that it tries to use __str__/tp_str
5518 on the object if the object is not a string or buffer. This
5519 makes unicode() behave like str() when applied to non-string/buffer
5522 - PyFile_WriteObject now passes Unicode objects to the file's write
5523 method. As a result, all file-like objects which may be the target
5524 of a print statement must support Unicode objects, i.e. they must
5525 at least convert them into ASCII strings.
5527 - Thread scheduling on Solaris should be improved; it is no longer
5528 necessary to insert a small sleep at the start of a thread in order
5529 to let other runnable threads be scheduled.
5534 - StringIO.StringIO instances and cStringIO.StringIO instances support
5535 read character buffer compatible objects for their .write() methods.
5536 These objects are converted to strings and then handled as such
5539 - The "email" package has been added. This is basically a port of the
5540 mimelib package <http://sf.net/projects/mimelib> with API changes
5541 and some implementations updated to use iterators and generators.
5543 - difflib.ndiff() and difflib.Differ.compare() are generators now. This
5544 restores the ability of Tools/scripts/ndiff.py to start producing output
5545 before the entire comparison is complete.
5547 - StringIO.StringIO instances and cStringIO.StringIO instances support
5548 iteration just like file objects (i.e. their .readline() method is
5549 called for each iteration until it returns an empty string).
5551 - The codecs module has grown four new helper APIs to access
5552 builtin codecs: getencoder(), getdecoder(), getreader(),
5555 - SimpleXMLRPCServer: a new module (based upon SimpleHTMLServer)
5556 simplifies writing XML RPC servers.
5558 - os.path.realpath(): a new function that returns the absolute pathname
5559 after interpretation of symbolic links. On non-Unix systems, this
5560 is an alias for os.path.abspath().
5562 - operator.indexOf() (PySequence_Index() in the C API) now works with any
5565 - smtplib now supports various authentication and security features of
5566 the SMTP protocol through the new login() and starttls() methods.
5568 - hmac: a new module implementing keyed hashing for message
5571 - mimetypes now recognizes more extensions and file types. At the
5572 same time, some mappings not sanctioned by IANA were removed.
5574 - The "compiler" package has been brought up to date to the state of
5575 Python 2.2 bytecode generation. It has also been promoted from a
5576 Tool to a standard library package. (Tools/compiler still exists as
5582 - Large file support (LFS) is now automatic when the platform supports
5583 it; no more manual configuration tweaks are needed. On Linux, at
5584 least, it's possible to have a system whose C library supports large
5585 files but whose kernel doesn't; in this case, large file support is
5586 still enabled but doesn't do you any good unless you upgrade your
5587 kernel or share your Python executable with another system whose
5588 kernel has large file support.
5590 - The configure script now supplies plausible defaults in a
5591 cross-compilation environment. This doesn't mean that the supplied
5592 values are always correct, or that cross-compilation now works
5593 flawlessly -- but it's a first step (and it shuts up most of
5594 autoconf's warnings about AC_TRY_RUN).
5596 - The Unix build is now a bit less chatty, courtesy of the parser
5597 generator. The build is completely silent (except for errors) when
5598 using "make -s", thanks to a -q option to setup.py.
5603 - The "structmember" API now supports some new flag bits to deny read
5604 and/or write access to attributes in restricted execution mode.
5609 - Compaq's iPAQ handheld, running the "familiar" Linux distribution
5610 (http://familiar.handhelds.org).
5615 - The "classic" standard tests, which work by comparing stdout to
5616 an expected-output file under Lib/test/output/, no longer stop at
5617 the first mismatch. Instead the test is run to completion, and a
5618 variant of ndiff-style comparison is used to report all differences.
5619 This is much easier to understand than the previous style of reporting.
5621 - The unittest-based standard tests now use regrtest's test_main()
5622 convention, instead of running as a side-effect of merely being
5623 imported. This allows these tests to be run in more natural and
5624 flexible ways as unittests, outside the regrtest framework.
5626 - regrtest.py is much better integrated with unittest and doctest now,
5627 especially in regard to reporting errors.
5632 - Large file support now also works for files > 4GB, on filesystems
5633 that support it (NTFS under Windows 2000). See "What's New in
5634 Python 2.2a3" for more detail.
5637 What's New in Python 2.2a3?
5638 ===========================
5640 *Release Date: 07-Sep-2001*
5645 - Conversion of long to float now raises OverflowError if the long is too
5646 big to represent as a C double.
5648 - The 3-argument builtin pow() no longer allows a third non-None argument
5649 if either of the first two arguments is a float, or if both are of
5650 integer types and the second argument is negative (in which latter case
5651 the arguments are converted to float, so this is really the same
5654 - The builtin dir() now returns more information, and sometimes much
5655 more, generally naming all attributes of an object, and all attributes
5656 reachable from the object via its class, and from its class's base
5657 classes, and so on from them too. Example: in 2.2a2, dir([]) returned
5658 an empty list. In 2.2a3,
5661 ['__add__', '__class__', '__contains__', '__delattr__', '__delitem__',
5662 '__eq__', '__ge__', '__getattr__', '__getitem__', '__getslice__',
5663 '__gt__', '__hash__', '__iadd__', '__imul__', '__init__', '__le__',
5664 '__len__', '__lt__', '__mul__', '__ne__', '__new__', '__repr__',
5665 '__rmul__', '__setattr__', '__setitem__', '__setslice__', '__str__',
5666 'append', 'count', 'extend', 'index', 'insert', 'pop', 'remove',
5669 dir(module) continues to return only the module's attributes, though.
5671 - Overflowing operations on plain ints now return a long int rather
5672 than raising OverflowError. This is a partial implementation of PEP
5673 237. You can use -Wdefault::OverflowWarning to enable a warning for
5674 this situation, and -Werror::OverflowWarning to revert to the old
5675 OverflowError exception.
5677 - A new command line option, -Q<arg>, is added to control run-time
5678 warnings for the use of classic division. (See PEP 238.) Possible
5679 values are -Qold, -Qwarn, -Qwarnall, and -Qnew. The default is
5680 -Qold, meaning the / operator has its classic meaning and no
5681 warnings are issued. Using -Qwarn issues a run-time warning about
5682 all uses of classic division for int and long arguments; -Qwarnall
5683 also warns about classic division for float and complex arguments
5684 (for use with fixdiv.py).
5685 [Note: the remainder of this item (preserved below) became
5686 obsolete in 2.2c1 -- -Qnew has global effect in 2.2] ::
5688 Using -Qnew is questionable; it turns on new division by default, but
5689 only in the __main__ module. You can usefully combine -Qwarn or
5690 -Qwarnall and -Qnew: this gives the __main__ module new division, and
5691 warns about classic division everywhere else.
5693 - Many built-in types can now be subclassed. This applies to int,
5694 long, float, str, unicode, and tuple. (The types complex, list and
5695 dictionary can also be subclassed; this was introduced earlier.)
5696 Note that restrictions apply when subclassing immutable built-in
5697 types: you can only affect the value of the instance by overloading
5698 __new__. You can add mutable attributes, and the subclass instances
5699 will have a __dict__ attribute, but you cannot change the "value"
5700 (as implemented by the base class) of an immutable subclass instance
5703 - The dictionary constructor now takes an optional argument, a
5704 mapping-like object, and initializes the dictionary from its
5707 - A new built-in type, super, has been added. This facilitates making
5708 "cooperative super calls" in a multiple inheritance setting. For an
5709 explanation, see http://www.python.org/2.2/descrintro.html#cooperation
5711 - A new built-in type, property, has been added. This enables the
5712 creation of "properties". These are attributes implemented by
5713 getter and setter functions (or only one of these for read-only or
5714 write-only attributes), without the need to override __getattr__.
5715 See http://www.python.org/2.2/descrintro.html#property
5717 - The syntax of floating-point and imaginary literals has been
5718 liberalized, to allow leading zeroes. Examples of literals now
5719 legal that were SyntaxErrors before:
5721 00.0 0e3 0100j 07.5 00000000000000000008.
5723 - An old tokenizer bug allowed floating point literals with an incomplete
5724 exponent, such as 1e and 3.1e-. Such literals now raise SyntaxError.
5729 - telnetlib includes symbolic names for the options, and support for
5730 setting an option negotiation callback. It also supports processing
5733 - The new C standard no longer requires that math libraries set errno to
5734 ERANGE on overflow. For platform libraries that exploit this new
5735 freedom, Python's overflow-checking was wholly broken. A new overflow-
5736 checking scheme attempts to repair that, but may not be reliable on all
5737 platforms (C doesn't seem to provide anything both useful and portable
5738 in this area anymore).
5740 - Asynchronous timeout actions are available through the new class
5743 - math.log and math.log10 now return sensible results for even huge
5744 long arguments. For example, math.log10(10 ** 10000) ~= 10000.0.
5746 - A new function, imp.lock_held(), returns 1 when the import lock is
5747 currently held. See the docs for the imp module.
5749 - pickle, cPickle and marshal on 32-bit platforms can now correctly read
5750 dumps containing ints written on platforms where Python ints are 8 bytes.
5751 When read on a box where Python ints are 4 bytes, such values are
5752 converted to Python longs.
5754 - In restricted execution mode (using the rexec module), unmarshalling
5755 code objects is no longer allowed. This plugs a security hole.
5757 - unittest.TestResult instances no longer store references to tracebacks
5758 generated by test failures. This prevents unexpected dangling references
5759 to objects that should be garbage collected between tests.
5764 - Tools/scripts/fixdiv.py has been added which can be used to fix
5765 division operators as per PEP 238.
5770 - If you are an adventurous person using Mac OS X you may want to look at
5771 Mac/OSX. There is a Makefile there that will build Python as a real Mac
5772 application, which can be used for experimenting with Carbon or Cocoa.
5773 Discussion of this on pythonmac-sig, please.
5778 - New function PyObject_Dir(obj), like Python __builtin__.dir(obj).
5780 - Note that PyLong_AsDouble can fail! This has always been true, but no
5781 callers checked for it. It's more likely to fail now, because overflow
5782 errors are properly detected now. The proper way to check::
5784 double x = PyLong_AsDouble(some_long_object);
5785 if (x == -1.0 && PyErr_Occurred()) {
5786 /* The conversion failed. */
5789 - The GC API has been changed. Extensions that use the old API will still
5790 compile but will not participate in GC. To upgrade an extension
5793 - rename Py_TPFLAGS_GC to PyTPFLAGS_HAVE_GC
5795 - use PyObject_GC_New or PyObject_GC_NewVar to allocate objects and
5796 PyObject_GC_Del to deallocate them
5798 - rename PyObject_GC_Init to PyObject_GC_Track and PyObject_GC_Fini
5799 to PyObject_GC_UnTrack
5801 - remove PyGC_HEAD_SIZE from object size calculations
5803 - remove calls to PyObject_AS_GC and PyObject_FROM_GC
5805 - Two new functions: PyString_FromFormat() and PyString_FromFormatV().
5806 These can be used safely to construct string objects from a
5807 sprintf-style format string (similar to the format string supported
5813 - Stephen Hansen contributed patches sufficient to get a clean compile
5814 under Borland C (Windows), but he reports problems running it and ran
5815 out of time to complete the port. Volunteers? Expect a MemoryError
5816 when importing the types module; this is probably shallow, and
5817 causing later failures too.
5825 - Large file support is now enabled on Win32 platforms as well as on
5826 Win64. This means that, for example, you can use f.tell() and f.seek()
5827 to manipulate files larger than 2 gigabytes (provided you have enough
5828 disk space, and are using a Windows filesystem that supports large
5829 partitions). Windows filesystem limits: FAT has a 2GB (gigabyte)
5830 filesize limit, and large file support makes no difference there.
5831 FAT32's limit is 4GB, and files >= 2GB are easier to use from Python now.
5832 NTFS has no practical limit on file size, and files of any size can be
5833 used from Python now.
5835 - The w9xpopen hack is now used on Windows NT and 2000 too when COMPSPEC
5836 points to command.com (patch from Brian Quinlan).
5839 What's New in Python 2.2a2?
5840 ===========================
5842 *Release Date: 22-Aug-2001*
5847 - Tim Peters developed a brand new Windows installer using Wise 8.1,
5848 generously donated to us by Wise Solutions.
5850 - configure supports a new option --enable-unicode, with the values
5851 ucs2 and ucs4 (new in 2.2a1). With --disable-unicode, the Unicode
5852 type and supporting code is completely removed from the interpreter.
5854 - A new configure option --enable-framework builds a Mac OS X framework,
5855 which "make frameworkinstall" will install. This provides a starting
5856 point for more mac-like functionality, join pythonmac-sig@python.org
5857 if you are interested in helping.
5859 - The NeXT platform is no longer supported.
5861 - The 'new' module is now statically linked.
5866 - The new Tools/scripts/cleanfuture.py can be used to automatically
5867 edit out obsolete future statements from Python source code. See
5868 the module docstring for details.
5873 - regrtest.py now knows which tests are expected to be skipped on some
5874 platforms, allowing to give clearer test result output. regrtest
5875 also has optional --use/-u switch to run normally disabled tests
5876 which require network access or consume significant disk resources.
5878 - Several new tests in the standard test suite, with special thanks to
5884 - The floor division operator // has been added as outlined in PEP
5885 238. The / operator still provides classic division (and will until
5886 Python 3.0) unless "from __future__ import division" is included, in
5887 which case the / operator will provide true division. The operator
5888 module provides truediv() and floordiv() functions. Augmented
5889 assignment variants are included, as are the equivalent overloadable
5890 methods and C API methods. See the PEP for a full discussion:
5891 <http://python.sf.net/peps/pep-0238.html>
5893 - Future statements are now effective in simulated interactive shells
5894 (like IDLE). This should "just work" by magic, but read Michael
5895 Hudson's "Future statements in simulated shells" PEP 264 for full
5896 details: <http://python.sf.net/peps/pep-0264.html>.
5898 - The type/class unification (PEP 252-253) was integrated into the
5899 trunk and is not so tentative any more (the exact specification of
5900 some features is still tentative). A lot of work has done on fixing
5901 bugs and adding robustness and features (performance still has to
5904 - Warnings about a mismatch in the Python API during extension import
5905 now use the Python warning framework (which makes it possible to
5906 write filters for these warnings).
5908 - A function's __dict__ (aka func_dict) will now always be a
5909 dictionary. It used to be possible to delete it or set it to None,
5910 but now both actions raise TypeErrors. It is still legal to set it
5911 to a dictionary object. Getting func.__dict__ before any attributes
5912 have been assigned now returns an empty dictionary instead of None.
5914 - A new command line option, -E, was added which disables the use of
5915 all environment variables, or at least those that are specifically
5916 significant to Python. Usually those have a name starting with
5917 "PYTHON". This was used to fix a problem where the tests fail if
5918 the user happens to have PYTHONHOME or PYTHONPATH pointing to an
5924 - New class Differ and new functions ndiff() and restore() in difflib.py.
5925 These package the algorithms used by the popular Tools/scripts/ndiff.py,
5926 for programmatic reuse.
5928 - New function xml.sax.saxutils.quoteattr(): Quote an XML attribute
5929 value using the minimal quoting required for the value; more
5930 reliable than using xml.sax.saxutils.escape() for attribute values.
5932 - Readline completion support for cmd.Cmd was added.
5934 - Calling os.tempnam() or os.tmpnam() generate RuntimeWarnings.
5936 - Added function threading.BoundedSemaphore()
5938 - Added Ka-Ping Yee's cgitb.py module.
5940 - The 'new' module now exposes the CO_xxx flags.
5942 - The gc module offers the get_referents function.
5950 - Two new APIs PyOS_snprintf() and PyOS_vsnprintf() were added
5951 which provide a cross-platform implementations for the
5952 relatively new snprintf()/vsnprintf() C lib APIs. In contrast to
5953 the standard sprintf() and vsprintf() C lib APIs, these versions
5954 apply bounds checking on the used buffer which enhances protection
5955 against buffer overruns.
5957 - Unicode APIs now use name mangling to assure that mixing interpreters
5958 and extensions using different Unicode widths is rendered next to
5959 impossible. Trying to import an incompatible Unicode-aware extension
5960 will result in an ImportError. Unicode extensions writers must make
5961 sure to check the Unicode width compatibility in their extensions by
5962 using at least one of the mangled Unicode APIs in the extension.
5964 - Two new flags METH_NOARGS and METH_O are available in method definition
5965 tables to simplify implementation of methods with no arguments and a
5966 single untyped argument. Calling such methods is more efficient than
5967 calling corresponding METH_VARARGS methods. METH_OLDARGS is now
5973 - "import module" now compiles module.pyw if it exists and nothing else
5977 What's New in Python 2.2a1?
5978 ===========================
5980 *Release date: 18-Jul-2001*
5985 - TENTATIVELY, a large amount of code implementing much of what's
5986 described in PEP 252 (Making Types Look More Like Classes) and PEP
5987 253 (Subtyping Built-in Types) was added. This will be released
5988 with Python 2.2a1. Documentation will be provided separately
5989 through http://www.python.org/2.2/. The purpose of releasing this
5990 with Python 2.2a1 is to test backwards compatibility. It is
5991 possible, though not likely, that a decision is made not to release
5992 this code as part of 2.2 final, if any serious backwards
5993 incompatibilities are found during alpha testing that cannot be
5996 - Generators were added; this is a new way to create an iterator (see
5997 below) using what looks like a simple function containing one or
5998 more 'yield' statements. See PEP 255. Since this adds a new
5999 keyword to the language, this feature must be enabled by including a
6000 future statement: "from __future__ import generators" (see PEP 236).
6001 Generators will become a standard feature in a future release
6002 (probably 2.3). Without this future statement, 'yield' remains an
6003 ordinary identifier, but a warning is issued each time it is used.
6004 (These warnings currently don't conform to the warnings framework of
6005 PEP 230; we intend to fix this in 2.2a2.)
6007 - The UTF-16 codec was modified to be more RFC compliant. It will now
6008 only remove BOM characters at the start of the string and then
6009 only if running in native mode (UTF-16-LE and -BE won't remove a
6010 leading BMO character).
6012 - Strings now have a new method .decode() to complement the already
6013 existing .encode() method. These two methods provide direct access
6014 to the corresponding decoders and encoders of the registered codecs.
6016 To enhance the usability of the .encode() method, the special
6017 casing of Unicode object return values was dropped (Unicode objects
6018 were auto-magically converted to string using the default encoding).
6020 Both methods will now return whatever the codec in charge of the
6021 requested encoding returns as object, e.g. Unicode codecs will
6022 return Unicode objects when decoding is requested ("äöü".decode("latin-1")
6023 will return u"äöü"). This enables codec writer to create codecs
6024 for various simple to use conversions.
6026 New codecs were added to demonstrate these new features (the .encode()
6027 and .decode() columns indicate the type of the returned objects):
6029 +---------+-----------+-----------+-----------------------------+
6030 |Name | .encode() | .decode() | Description |
6031 +=========+===========+===========+=============================+
6032 |uu | string | string | UU codec (e.g. for email) |
6033 +---------+-----------+-----------+-----------------------------+
6034 |base64 | string | string | base64 codec |
6035 +---------+-----------+-----------+-----------------------------+
6036 |quopri | string | string | quoted-printable codec |
6037 +---------+-----------+-----------+-----------------------------+
6038 |zlib | string | string | zlib compression |
6039 +---------+-----------+-----------+-----------------------------+
6040 |hex | string | string | 2-byte hex codec |
6041 +---------+-----------+-----------+-----------------------------+
6042 |rot-13 | string | Unicode | ROT-13 Unicode charmap codec|
6043 +---------+-----------+-----------+-----------------------------+
6045 - Some operating systems now support the concept of a default Unicode
6046 encoding for file system operations. Notably, Windows supports 'mbcs'
6047 as the default. The Macintosh will also adopt this concept in the medium
6048 term, although the default encoding for that platform will be other than
6051 On operating system that support non-ASCII filenames, it is common for
6052 functions that return filenames (such as os.listdir()) to return Python
6053 string objects pre-encoded using the default file system encoding for
6054 the platform. As this encoding is likely to be different from Python's
6055 default encoding, converting this name to a Unicode object before passing
6056 it back to the Operating System would result in a Unicode error, as Python
6057 would attempt to use its default encoding (generally ASCII) rather than
6058 the default encoding for the file system.
6060 In general, this change simply removes surprises when working with
6061 Unicode and the file system, making these operations work as you expect,
6062 increasing the transparency of Unicode objects in this context.
6063 See [????] for more details, including examples.
6065 - Float (and complex) literals in source code were evaluated to full
6066 precision only when running from a .py file; the same code loaded from a
6067 .pyc (or .pyo) file could suffer numeric differences starting at about the
6068 12th significant decimal digit. For example, on a machine with IEEE-754
6069 floating arithmetic,
6071 x = 9007199254740992.0
6074 printed 9007199254740992 if run directly from .py, but 9007199254740000
6075 if from a compiled (.pyc or .pyo) file. This was due to marshal using
6076 str(float) instead of repr(float) when building code objects. marshal
6077 now uses repr(float) instead, which should reproduce floats to full
6078 machine precision (assuming the platform C float<->string I/O conversion
6079 functions are of good quality).
6081 This may cause floating-point results to change in some cases, and
6082 usually for the better, but may also cause numerically unstable
6083 algorithms to break.
6085 - The implementation of dicts suffers fewer collisions, which has speed
6086 benefits. However, the order in which dict entries appear in dict.keys(),
6087 dict.values() and dict.items() may differ from previous releases for a
6088 given dict. Nothing is defined about this order, so no program should
6089 rely on it. Nevertheless, it's easy to write test cases that rely on the
6090 order by accident, typically because of printing the str() or repr() of a
6091 dict to an "expected results" file. See Lib/test/test_support.py's new
6092 sortdict(dict) function for a simple way to display a dict in sorted
6095 - Many other small changes to dicts were made, resulting in faster
6096 operation along the most common code paths.
6098 - Dictionary objects now support the "in" operator: "x in dict" means
6099 the same as dict.has_key(x).
6101 - The update() method of dictionaries now accepts generic mapping
6102 objects. Specifically the argument object must support the .keys()
6103 and __getitem__() methods. This allows you to say, for example,
6104 {}.update(UserDict())
6106 - Iterators were added; this is a generalized way of providing values
6107 to a for loop. See PEP 234. There's a new built-in function iter()
6108 to return an iterator. There's a new protocol to get the next value
6109 from an iterator using the next() method (in Python) or the
6110 tp_iternext slot (in C). There's a new protocol to get iterators
6111 using the __iter__() method (in Python) or the tp_iter slot (in C).
6112 Iterating (i.e. a for loop) over a dictionary generates its keys.
6113 Iterating over a file generates its lines.
6115 - The following functions were generalized to work nicely with iterator
6118 map(), filter(), reduce(), zip()
6119 list(), tuple() (PySequence_Tuple() and PySequence_Fast() in C API)
6121 join() method of strings
6122 extend() method of lists
6123 'x in y' and 'x not in y' (PySequence_Contains() in C API)
6124 operator.countOf() (PySequence_Count() in C API)
6125 right-hand side of assignment statements with multiple targets, such as ::
6126 x, y, z = some_iterable_object_returning_exactly_3_values
6128 - Accessing module attributes is significantly faster (for example,
6129 random.random or os.path or yourPythonModule.yourAttribute).
6131 - Comparing dictionary objects via == and != is faster, and now works even
6132 if the keys and values don't support comparisons other than ==.
6134 - Comparing dictionaries in ways other than == and != is slower: there were
6135 insecurities in the dict comparison implementation that could cause Python
6136 to crash if the element comparison routines for the dict keys and/or
6137 values mutated the dicts. Making the code bulletproof slowed it down.
6139 - Collisions in dicts are resolved via a new approach, which can help
6140 dramatically in bad cases. For example, looking up every key in a dict
6141 d with d.keys() == [i << 16 for i in range(20000)] is approximately 500x
6142 faster now. Thanks to Christian Tismer for pointing out the cause and
6143 the nature of an effective cure (last December! better late than never).
6145 - repr() is much faster for large containers (dict, list, tuple).
6151 - The constants ascii_letters, ascii_lowercase. and ascii_uppercase
6152 were added to the string module. These a locale-independent
6153 constants, unlike letters, lowercase, and uppercase. These are now
6154 use in appropriate locations in the standard library.
6156 - The flags used in dlopen calls can now be configured using
6157 sys.setdlopenflags and queried using sys.getdlopenflags.
6159 - Fredrik Lundh's xmlrpclib is now a standard library module. This
6160 provides full client-side XML-RPC support. In addition,
6161 Demo/xmlrpc/ contains two server frameworks (one SocketServer-based,
6162 one asyncore-based). Thanks to Eric Raymond for the documentation.
6164 - The xrange() object is simplified: it no longer supports slicing,
6165 repetition, comparisons, efficient 'in' checking, the tolist()
6166 method, or the start, stop and step attributes. See PEP 260.
6168 - A new function fnmatch.filter to filter lists of file names was added.
6170 - calendar.py uses month and day names based on the current locale.
6172 - strop is now *really* obsolete (this was announced before with 1.6),
6173 and issues DeprecationWarning when used (except for the four items
6174 that are still imported into string.py).
6176 - Cookie.py now sorts key+value pairs by key in output strings.
6178 - pprint.isrecursive(object) didn't correctly identify recursive objects.
6181 - pprint functions now much faster for large containers (tuple, list, dict).
6183 - New 'q' and 'Q' format codes in the struct module, corresponding to C
6184 types "long long" and "unsigned long long" (on Windows, __int64). In
6185 native mode, these can be used only when the platform C compiler supports
6186 these types (when HAVE_LONG_LONG is #define'd by the Python config
6187 process), and then they inherit the sizes and alignments of the C types.
6188 In standard mode, 'q' and 'Q' are supported on all platforms, and are
6189 8-byte integral types.
6191 - The site module installs a new built-in function 'help' that invokes
6192 pydoc.help. It must be invoked as 'help()'; when invoked as 'help',
6193 it displays a message reminding the user to use 'help()' or
6199 - New test_mutants.py runs dict comparisons where the key and value
6200 comparison operators mutate the dicts randomly during comparison. This
6201 rapidly causes Python to crash under earlier releases (not for the faint
6202 of heart: it can also cause Win9x to freeze or reboot!).
6204 - New test_pprint.py verifies that pprint.isrecursive() and
6205 pprint.isreadable() return sensible results. Also verifies that simple
6206 cases produce correct output.
6211 - Removed the unused last_is_sticky argument from the internal
6212 _PyTuple_Resize(). If this affects you, you were cheating.
6216 **(For information about older versions, consult the HISTORY file.)**