Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 1 | **************************** |
| 2 | What's New In Python 3.3 |
| 3 | **************************** |
| 4 | |
| 5 | :Author: Raymond Hettinger |
| 6 | :Release: |release| |
| 7 | :Date: |today| |
| 8 | |
Éric Araujo | b07b97f | 2011-10-05 01:03:34 +0200 | [diff] [blame] | 9 | .. Rules for maintenance: |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 10 | |
| 11 | * Anyone can add text to this document. Do not spend very much time |
| 12 | on the wording of your changes, because your text will probably |
| 13 | get rewritten to some degree. |
| 14 | |
| 15 | * The maintainer will go through Misc/NEWS periodically and add |
| 16 | changes; it's therefore more important to add your changes to |
| 17 | Misc/NEWS than to this file. |
| 18 | |
| 19 | * This is not a complete list of every single change; completeness |
| 20 | is the purpose of Misc/NEWS. Some changes I consider too small |
| 21 | or esoteric to include. If such a change is added to the text, |
| 22 | I'll just remove it. (This is another reason you shouldn't spend |
| 23 | too much time on writing your addition.) |
| 24 | |
| 25 | * If you want to draw your new text to the attention of the |
| 26 | maintainer, add 'XXX' to the beginning of the paragraph or |
| 27 | section. |
| 28 | |
| 29 | * It's OK to just add a fragmentary note about a change. For |
| 30 | example: "XXX Describe the transmogrify() function added to the |
| 31 | socket module." The maintainer will research the change and |
| 32 | write the necessary text. |
| 33 | |
| 34 | * You can comment out your additions if you like, but it's not |
| 35 | necessary (especially when a final release is some months away). |
| 36 | |
| 37 | * Credit the author of a patch or bugfix. Just the name is |
| 38 | sufficient; the e-mail address isn't necessary. |
| 39 | |
| 40 | * It's helpful to add the bug/patch number as a comment: |
| 41 | |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 42 | XXX Describe the transmogrify() function added to the socket |
| 43 | module. |
Éric Araujo | b07b97f | 2011-10-05 01:03:34 +0200 | [diff] [blame] | 44 | (Contributed by P.Y. Developer in :issue:`12345`.) |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 45 | |
Éric Araujo | b07b97f | 2011-10-05 01:03:34 +0200 | [diff] [blame] | 46 | This saves the maintainer the effort of going through the Mercurial log |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 47 | when researching a change. |
| 48 | |
| 49 | This article explains the new features in Python 3.3, compared to 3.2. |
| 50 | |
Nick Coghlan | b47b539 | 2012-05-26 01:31:25 +1000 | [diff] [blame] | 51 | .. note:: Alpha users should be aware that this document is currently in |
| 52 | draft form. It will be updated substantially as Python 3.3 moves towards |
| 53 | release, so it's worth checking back even after reading earlier versions. |
| 54 | |
| 55 | |
| 56 | New packaging infrastructure |
| 57 | ============================ |
| 58 | |
| 59 | The standard library's packaging infrastructure has been updated to adopt |
| 60 | some of the features developed by the wider community. |
| 61 | |
| 62 | * the :mod:`packaging` package and ``pysetup`` script (inspired by |
| 63 | ``setuptools``, ``distribute``, ``distutil2`` and ``pip``) |
| 64 | * the :mod:`venv` module and ``pyvenv`` script (inspired by ``virtualenv``) |
| 65 | (Note: at time of writing, :pep:`405` is accepted, but not yet implemented) |
| 66 | * native support for package directories that don't require ``__init__.py`` |
| 67 | marker files and can automatically span multiple path segments |
| 68 | (inspired by various third party approaches to namespace packages, as |
| 69 | described in :pep:`420`) |
| 70 | |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 71 | |
Nick Coghlan | 98e2070 | 2012-03-06 21:50:13 +1000 | [diff] [blame] | 72 | .. pep-3118-update: |
| 73 | |
Stefan Krah | 9a2d99e | 2012-02-25 12:24:21 +0100 | [diff] [blame] | 74 | PEP 3118: New memoryview implementation and buffer protocol documentation |
| 75 | ========================================================================= |
| 76 | |
| 77 | :issue:`10181` - memoryview bug fixes and features. |
| 78 | Written by Stefan Krah. |
| 79 | |
| 80 | The new memoryview implementation comprehensively fixes all ownership and |
| 81 | lifetime issues of dynamically allocated fields in the Py_buffer struct |
| 82 | that led to multiple crash reports. Additionally, several functions that |
| 83 | crashed or returned incorrect results for non-contiguous or multi-dimensional |
| 84 | input have been fixed. |
| 85 | |
| 86 | The memoryview object now has a PEP-3118 compliant getbufferproc() |
| 87 | that checks the consumer's request type. Many new features have been |
| 88 | added, most of them work in full generality for non-contiguous arrays |
| 89 | and arrays with suboffsets. |
| 90 | |
| 91 | The documentation has been updated, clearly spelling out responsibilities |
| 92 | for both exporters and consumers. Buffer request flags are grouped into |
| 93 | basic and compound flags. The memory layout of non-contiguous and |
| 94 | multi-dimensional NumPy-style arrays is explained. |
| 95 | |
| 96 | Features |
| 97 | -------- |
| 98 | |
| 99 | * All native single character format specifiers in struct module syntax |
| 100 | (optionally prefixed with '@') are now supported. |
| 101 | |
| 102 | * With some restrictions, the cast() method allows changing of format and |
| 103 | shape of C-contiguous arrays. |
| 104 | |
| 105 | * Multi-dimensional list representations are supported for any array type. |
| 106 | |
| 107 | * Multi-dimensional comparisons are supported for any array type. |
| 108 | |
| 109 | * All array types are hashable if the exporting object is hashable |
Nick Coghlan | 98e2070 | 2012-03-06 21:50:13 +1000 | [diff] [blame] | 110 | and the view is read-only. (Contributed by Antoine Pitrou in |
| 111 | :issue:`13411`) |
| 112 | |
Stefan Krah | 9a2d99e | 2012-02-25 12:24:21 +0100 | [diff] [blame] | 113 | |
| 114 | * Arbitrary slicing of any 1-D arrays type is supported. For example, it |
| 115 | is now possible to reverse a memoryview in O(1) by using a negative step. |
| 116 | |
| 117 | API changes |
| 118 | ----------- |
| 119 | |
| 120 | * The maximum number of dimensions is officially limited to 64. |
| 121 | |
| 122 | * The representation of empty shape, strides and suboffsets is now |
| 123 | an empty tuple instead of None. |
| 124 | |
| 125 | * Accessing a memoryview element with format 'B' (unsigned bytes) |
| 126 | now returns an integer (in accordance with the struct module syntax). |
| 127 | For returning a bytes object the view must be cast to 'c' first. |
| 128 | |
Stefan Krah | 54c3203 | 2012-02-29 17:47:21 +0100 | [diff] [blame] | 129 | * For further changes see `Build and C API Changes`_ and `Porting C code`_ . |
Stefan Krah | 9a2d99e | 2012-02-25 12:24:21 +0100 | [diff] [blame] | 130 | |
Antoine Pitrou | 037ffbf | 2011-10-24 00:25:41 +0200 | [diff] [blame] | 131 | .. _pep-393: |
| 132 | |
Ezio Melotti | 48a2f8f | 2011-09-29 00:18:19 +0300 | [diff] [blame] | 133 | PEP 393: Flexible String Representation |
| 134 | ======================================= |
| 135 | |
Antoine Pitrou | fd9b416 | 2011-10-24 00:14:43 +0200 | [diff] [blame] | 136 | The Unicode string type is changed to support multiple internal |
| 137 | representations, depending on the character with the largest Unicode ordinal |
| 138 | (1, 2, or 4 bytes) in the represented string. This allows a space-efficient |
| 139 | representation in common cases, but gives access to full UCS-4 on all |
| 140 | systems. For compatibility with existing APIs, several representations may |
| 141 | exist in parallel; over time, this compatibility should be phased out. |
Ezio Melotti | 397546a | 2011-09-29 08:34:36 +0300 | [diff] [blame] | 142 | |
Antoine Pitrou | fd9b416 | 2011-10-24 00:14:43 +0200 | [diff] [blame] | 143 | On the Python side, there should be no downside to this change. |
Ezio Melotti | 397546a | 2011-09-29 08:34:36 +0300 | [diff] [blame] | 144 | |
Antoine Pitrou | fd9b416 | 2011-10-24 00:14:43 +0200 | [diff] [blame] | 145 | On the C API side, PEP 393 is fully backward compatible. The legacy API |
| 146 | should remain available at least five years. Applications using the legacy |
| 147 | API will not fully benefit of the memory reduction, or - worse - may use |
| 148 | a bit more memory, because Python may have to maintain two versions of each |
| 149 | string (in the legacy format and in the new efficient storage). |
| 150 | |
Antoine Pitrou | 0599b5b | 2011-11-29 22:45:07 +0100 | [diff] [blame] | 151 | Functionality |
| 152 | ------------- |
| 153 | |
Antoine Pitrou | fd9b416 | 2011-10-24 00:14:43 +0200 | [diff] [blame] | 154 | Changes introduced by :pep:`393` are the following: |
Ezio Melotti | 48a2f8f | 2011-09-29 00:18:19 +0300 | [diff] [blame] | 155 | |
Ezio Melotti | 397546a | 2011-09-29 08:34:36 +0300 | [diff] [blame] | 156 | * Python now always supports the full range of Unicode codepoints, including |
| 157 | non-BMP ones (i.e. from ``U+0000`` to ``U+10FFFF``). The distinction between |
| 158 | narrow and wide builds no longer exists and Python now behaves like a wide |
Antoine Pitrou | fd9b416 | 2011-10-24 00:14:43 +0200 | [diff] [blame] | 159 | build, even under Windows. |
Ezio Melotti | 397546a | 2011-09-29 08:34:36 +0300 | [diff] [blame] | 160 | |
Antoine Pitrou | fd9b416 | 2011-10-24 00:14:43 +0200 | [diff] [blame] | 161 | * With the death of narrow builds, the problems specific to narrow builds have |
| 162 | also been fixed, for example: |
Ezio Melotti | 397546a | 2011-09-29 08:34:36 +0300 | [diff] [blame] | 163 | |
| 164 | * :func:`len` now always returns 1 for non-BMP characters, |
| 165 | so ``len('\U0010FFFF') == 1``; |
| 166 | |
| 167 | * surrogate pairs are not recombined in string literals, |
| 168 | so ``'\uDBFF\uDFFF' != '\U0010FFFF'``; |
| 169 | |
Antoine Pitrou | fd9b416 | 2011-10-24 00:14:43 +0200 | [diff] [blame] | 170 | * indexing or slicing non-BMP characters returns the expected value, |
Ezio Melotti | 397546a | 2011-09-29 08:34:36 +0300 | [diff] [blame] | 171 | so ``'\U0010FFFF'[0]`` now returns ``'\U0010FFFF'`` and not ``'\uDBFF'``; |
| 172 | |
Antoine Pitrou | d136aec | 2011-11-17 01:48:06 +0100 | [diff] [blame] | 173 | * all other functions in the standard library now correctly handle |
Antoine Pitrou | fd9b416 | 2011-10-24 00:14:43 +0200 | [diff] [blame] | 174 | non-BMP codepoints. |
Ezio Melotti | 397546a | 2011-09-29 08:34:36 +0300 | [diff] [blame] | 175 | |
Ezio Melotti | 48a2f8f | 2011-09-29 00:18:19 +0300 | [diff] [blame] | 176 | * The value of :data:`sys.maxunicode` is now always ``1114111`` (``0x10FFFF`` |
| 177 | in hexadecimal). The :c:func:`PyUnicode_GetMax` function still returns |
| 178 | either ``0xFFFF`` or ``0x10FFFF`` for backward compatibility, and it should |
| 179 | not be used with the new Unicode API (see :issue:`13054`). |
| 180 | |
Ezio Melotti | 397546a | 2011-09-29 08:34:36 +0300 | [diff] [blame] | 181 | * The :file:`./configure` flag ``--with-wide-unicode`` has been removed. |
Victor Stinner | 7d637ab | 2011-09-29 02:56:16 +0200 | [diff] [blame] | 182 | |
Antoine Pitrou | 0599b5b | 2011-11-29 22:45:07 +0100 | [diff] [blame] | 183 | Performance and resource usage |
| 184 | ------------------------------ |
| 185 | |
| 186 | The storage of Unicode strings now depends on the highest codepoint in the string: |
| 187 | |
| 188 | * pure ASCII and Latin1 strings (``U+0000-U+00FF``) use 1 byte per codepoint; |
| 189 | |
| 190 | * BMP strings (``U+0000-U+FFFF``) use 2 bytes per codepoint; |
| 191 | |
| 192 | * non-BMP strings (``U+10000-U+10FFFF``) use 4 bytes per codepoint. |
| 193 | |
Martin v. Löwis | de157cc | 2012-03-06 08:42:17 +0100 | [diff] [blame] | 194 | The net effect is that for most applications, memory usage of string |
| 195 | storage should decrease significantly - especially compared to former |
| 196 | wide unicode builds - as, in many cases, strings will be pure ASCII |
| 197 | even in international contexts (because many strings store non-human |
| 198 | language data, such as XML fragments, HTTP headers, JSON-encoded data, |
| 199 | etc.). We also hope that it will, for the same reasons, increase CPU |
| 200 | cache efficiency on non-trivial applications. The memory usage of |
| 201 | Python 3.3 is two to three times smaller than Python 3.2, and a little |
| 202 | bit better than Python 2.7, on a Django benchmark (see the PEP for |
| 203 | details). |
Antoine Pitrou | 0599b5b | 2011-11-29 22:45:07 +0100 | [diff] [blame] | 204 | |
Éric Araujo | b07b97f | 2011-10-05 01:03:34 +0200 | [diff] [blame] | 205 | |
Victor Stinner | a1bf298 | 2011-10-12 20:35:02 +0200 | [diff] [blame] | 206 | PEP 3151: Reworking the OS and IO exception hierarchy |
| 207 | ===================================================== |
| 208 | |
| 209 | :pep:`3151` - Reworking the OS and IO exception hierarchy |
Antoine Pitrou | 01fd26c | 2011-10-24 00:07:02 +0200 | [diff] [blame] | 210 | PEP written and implemented by Antoine Pitrou. |
Victor Stinner | a1bf298 | 2011-10-12 20:35:02 +0200 | [diff] [blame] | 211 | |
Antoine Pitrou | 01fd26c | 2011-10-24 00:07:02 +0200 | [diff] [blame] | 212 | The hierarchy of exceptions raised by operating system errors is now both |
| 213 | simplified and finer-grained. |
Victor Stinner | a1bf298 | 2011-10-12 20:35:02 +0200 | [diff] [blame] | 214 | |
Antoine Pitrou | 01fd26c | 2011-10-24 00:07:02 +0200 | [diff] [blame] | 215 | You don't have to worry anymore about choosing the appropriate exception |
| 216 | type between :exc:`OSError`, :exc:`IOError`, :exc:`EnvironmentError`, |
| 217 | :exc:`WindowsError`, :exc:`mmap.error`, :exc:`socket.error` or |
| 218 | :exc:`select.error`. All these exception types are now only one: |
| 219 | :exc:`OSError`. The other names are kept as aliases for compatibility |
| 220 | reasons. |
Victor Stinner | a1bf298 | 2011-10-12 20:35:02 +0200 | [diff] [blame] | 221 | |
Antoine Pitrou | 01fd26c | 2011-10-24 00:07:02 +0200 | [diff] [blame] | 222 | Also, it is now easier to catch a specific error condition. Instead of |
| 223 | inspecting the ``errno`` attribute (or ``args[0]``) for a particular |
| 224 | constant from the :mod:`errno` module, you can catch the adequate |
| 225 | :exc:`OSError` subclass. The available subclasses are the following: |
Victor Stinner | a1bf298 | 2011-10-12 20:35:02 +0200 | [diff] [blame] | 226 | |
Antoine Pitrou | 01fd26c | 2011-10-24 00:07:02 +0200 | [diff] [blame] | 227 | * :exc:`BlockingIOError` |
| 228 | * :exc:`ChildProcessError` |
| 229 | * :exc:`ConnectionError` |
| 230 | * :exc:`FileExistsError` |
| 231 | * :exc:`FileNotFoundError` |
| 232 | * :exc:`InterruptedError` |
| 233 | * :exc:`IsADirectoryError` |
| 234 | * :exc:`NotADirectoryError` |
| 235 | * :exc:`PermissionError` |
| 236 | * :exc:`ProcessLookupError` |
| 237 | * :exc:`TimeoutError` |
Victor Stinner | a1bf298 | 2011-10-12 20:35:02 +0200 | [diff] [blame] | 238 | |
Antoine Pitrou | 01fd26c | 2011-10-24 00:07:02 +0200 | [diff] [blame] | 239 | And the :exc:`ConnectionError` itself has finer-grained subclasses: |
Victor Stinner | a1bf298 | 2011-10-12 20:35:02 +0200 | [diff] [blame] | 240 | |
Antoine Pitrou | 01fd26c | 2011-10-24 00:07:02 +0200 | [diff] [blame] | 241 | * :exc:`BrokenPipeError` |
| 242 | * :exc:`ConnectionAbortedError` |
| 243 | * :exc:`ConnectionRefusedError` |
| 244 | * :exc:`ConnectionResetError` |
Victor Stinner | a1bf298 | 2011-10-12 20:35:02 +0200 | [diff] [blame] | 245 | |
| 246 | Thanks to the new exceptions, common usages of the :mod:`errno` can now be |
Antoine Pitrou | 01fd26c | 2011-10-24 00:07:02 +0200 | [diff] [blame] | 247 | avoided. For example, the following code written for Python 3.2:: |
Victor Stinner | a1bf298 | 2011-10-12 20:35:02 +0200 | [diff] [blame] | 248 | |
| 249 | from errno import ENOENT, EACCES, EPERM |
| 250 | |
| 251 | try: |
| 252 | with open("document.txt") as f: |
| 253 | content = f.read() |
| 254 | except IOError as err: |
| 255 | if err.errno == ENOENT: |
| 256 | print("document.txt file is missing") |
| 257 | elif err.errno in (EACCES, EPERM): |
| 258 | print("You are not allowed to read document.txt") |
| 259 | else: |
| 260 | raise |
| 261 | |
Antoine Pitrou | 01fd26c | 2011-10-24 00:07:02 +0200 | [diff] [blame] | 262 | can now be written without the :mod:`errno` import and without manual |
| 263 | inspection of exception attributes:: |
Victor Stinner | a1bf298 | 2011-10-12 20:35:02 +0200 | [diff] [blame] | 264 | |
| 265 | try: |
| 266 | with open("document.txt") as f: |
| 267 | content = f.read() |
| 268 | except FileNotFoundError: |
| 269 | print("document.txt file is missing") |
| 270 | except PermissionError: |
| 271 | print("You are not allowed to read document.txt") |
| 272 | |
| 273 | |
Nick Coghlan | 1f7ce62 | 2012-01-13 21:43:40 +1000 | [diff] [blame] | 274 | PEP 380: Syntax for Delegating to a Subgenerator |
| 275 | ================================================ |
| 276 | |
Nick Coghlan | ab7bf21 | 2012-02-26 17:49:52 +1000 | [diff] [blame] | 277 | :pep:`380` - Syntax for Delegating to a Subgenerator |
| 278 | PEP written by Greg Ewing. |
| 279 | |
Nick Coghlan | 1f7ce62 | 2012-01-13 21:43:40 +1000 | [diff] [blame] | 280 | PEP 380 adds the ``yield from`` expression, allowing a generator to delegate |
| 281 | part of its operations to another generator. This allows a section of code |
| 282 | containing 'yield' to be factored out and placed in another generator. |
| 283 | Additionally, the subgenerator is allowed to return with a value, and the |
| 284 | value is made available to the delegating generator. |
Nick Coghlan | b9b281b | 2012-03-06 22:31:12 +1000 | [diff] [blame] | 285 | |
Nick Coghlan | 1f7ce62 | 2012-01-13 21:43:40 +1000 | [diff] [blame] | 286 | While designed primarily for use in delegating to a subgenerator, the ``yield |
| 287 | from`` expression actually allows delegation to arbitrary subiterators. |
| 288 | |
Nick Coghlan | b9b281b | 2012-03-06 22:31:12 +1000 | [diff] [blame] | 289 | For simple iterators, ``yield from iterable`` is essentially just a shortened |
| 290 | form of ``for item in iterable: yield item``:: |
| 291 | |
| 292 | >>> def g(x): |
| 293 | ... yield from range(x, 0, -1) |
| 294 | ... yield from range(x) |
| 295 | ... |
| 296 | >>> list(g(5)) |
| 297 | [5, 4, 3, 2, 1, 0, 1, 2, 3, 4] |
| 298 | |
| 299 | However, unlike an ordinary loop, ``yield from`` allows subgenerators to |
| 300 | receive sent and thrown values directly from the calling scope, and |
| 301 | return a final value to the outer generator:: |
| 302 | |
| 303 | >>> def accumulate(start=0): |
| 304 | ... tally = start |
| 305 | ... while 1: |
| 306 | ... next = yield |
| 307 | ... if next is None: |
| 308 | ... return tally |
| 309 | ... tally += next |
| 310 | ... |
| 311 | >>> def gather_tallies(tallies, start=0): |
| 312 | ... while 1: |
| 313 | ... tally = yield from accumulate() |
| 314 | ... tallies.append(tally) |
| 315 | ... |
| 316 | >>> tallies = [] |
| 317 | >>> acc = gather_tallies(tallies) |
| 318 | >>> next(acc) # Ensure the accumulator is ready to accept values |
| 319 | >>> for i in range(10): |
| 320 | ... acc.send(i) |
| 321 | ... |
| 322 | >>> acc.send(None) # Finish the first tally |
| 323 | >>> for i in range(5): |
| 324 | ... acc.send(i) |
| 325 | ... |
| 326 | >>> acc.send(None) # Finish the second tally |
| 327 | >>> tallies |
| 328 | [45, 10] |
| 329 | |
| 330 | The main principle driving this change is to allow even generators that are |
| 331 | designed to be used with the ``send`` and ``throw`` methods to be split into |
| 332 | multiple subgenerators as easily as a single large function can be split into |
| 333 | multiple subfunctions. |
| 334 | |
Nick Coghlan | 1f7ce62 | 2012-01-13 21:43:40 +1000 | [diff] [blame] | 335 | (Implementation by Greg Ewing, integrated into 3.3 by Renaud Blanch, Ryan |
| 336 | Kelly and Nick Coghlan, documentation by Zbigniew Jędrzejewski-Szmek and |
| 337 | Nick Coghlan) |
| 338 | |
| 339 | |
Nick Coghlan | ab7bf21 | 2012-02-26 17:49:52 +1000 | [diff] [blame] | 340 | PEP 409: Suppressing exception context |
| 341 | ====================================== |
| 342 | |
| 343 | :pep:`409` - Suppressing exception context |
| 344 | PEP written by Ethan Furman, implemented by Ethan Furman and Nick Coghlan. |
| 345 | |
| 346 | PEP 409 introduces new syntax that allows the display of the chained |
| 347 | exception context to be disabled. This allows cleaner error messages in |
| 348 | applications that convert between exception types:: |
| 349 | |
| 350 | >>> class D: |
| 351 | ... def __init__(self, extra): |
| 352 | ... self._extra_attributes = extra |
| 353 | ... def __getattr__(self, attr): |
| 354 | ... try: |
| 355 | ... return self._extra_attributes[attr] |
| 356 | ... except KeyError: |
| 357 | ... raise AttributeError(attr) from None |
| 358 | ... |
| 359 | >>> D({}).x |
| 360 | Traceback (most recent call last): |
| 361 | File "<stdin>", line 1, in <module> |
| 362 | File "<stdin>", line 8, in __getattr__ |
| 363 | AttributeError: x |
| 364 | |
| 365 | Without the ``from None`` suffix to suppress the cause, the original |
| 366 | exception would be displayed by default:: |
| 367 | |
| 368 | >>> class C: |
| 369 | ... def __init__(self, extra): |
| 370 | ... self._extra_attributes = extra |
| 371 | ... def __getattr__(self, attr): |
| 372 | ... try: |
| 373 | ... return self._extra_attributes[attr] |
| 374 | ... except KeyError: |
| 375 | ... raise AttributeError(attr) |
| 376 | ... |
| 377 | >>> C({}).x |
| 378 | Traceback (most recent call last): |
| 379 | File "<stdin>", line 6, in __getattr__ |
| 380 | KeyError: 'x' |
| 381 | |
| 382 | During handling of the above exception, another exception occurred: |
| 383 | |
| 384 | Traceback (most recent call last): |
| 385 | File "<stdin>", line 1, in <module> |
| 386 | File "<stdin>", line 8, in __getattr__ |
| 387 | AttributeError: x |
| 388 | |
| 389 | No debugging capability is lost, as the original exception context remains |
| 390 | available if needed (for example, if an intervening library has incorrectly |
| 391 | suppressed valuable underlying details):: |
| 392 | |
| 393 | >>> try: |
| 394 | ... D({}).x |
| 395 | ... except AttributeError as exc: |
| 396 | ... print(repr(exc.__context__)) |
| 397 | ... |
| 398 | KeyError('x',) |
| 399 | |
| 400 | |
Nick Coghlan | 98e2070 | 2012-03-06 21:50:13 +1000 | [diff] [blame] | 401 | PEP 414: Explicit Unicode literals |
| 402 | ====================================== |
| 403 | |
| 404 | :pep:`414` - Explicit Unicode literals |
| 405 | PEP written by Armin Ronacher. |
| 406 | |
| 407 | To ease the transition from Python 2 for Unicode aware Python applications |
| 408 | that make heavy use of Unicode literals, Python 3.3 once again supports the |
| 409 | "``u``" prefix for string literals. This prefix has no semantic significance |
| 410 | in Python 3, it is provided solely to reduce the number of purely mechanical |
| 411 | changes in migrating to Python 3, making it easier for developers to focus on |
| 412 | the more significant semantic changes (such as the stricter default |
| 413 | separation of binary and text data). |
| 414 | |
| 415 | |
Antoine Pitrou | 6bbd76b | 2011-11-25 19:10:05 +0100 | [diff] [blame] | 416 | PEP 3155: Qualified name for classes and functions |
| 417 | ================================================== |
| 418 | |
| 419 | :pep:`3155` - Qualified name for classes and functions |
| 420 | PEP written and implemented by Antoine Pitrou. |
| 421 | |
| 422 | Functions and class objects have a new ``__qualname__`` attribute representing |
| 423 | the "path" from the module top-level to their definition. For global functions |
| 424 | and classes, this is the same as ``__name__``. For other functions and classes, |
| 425 | it provides better information about where they were actually defined, and |
| 426 | how they might be accessible from the global scope. |
| 427 | |
| 428 | Example with (non-bound) methods:: |
Nick Coghlan | 2dfe6b0 | 2012-01-14 14:19:49 +1000 | [diff] [blame] | 429 | |
Antoine Pitrou | 6bbd76b | 2011-11-25 19:10:05 +0100 | [diff] [blame] | 430 | >>> class C: |
| 431 | ... def meth(self): |
| 432 | ... pass |
| 433 | >>> C.meth.__name__ |
| 434 | 'meth' |
| 435 | >>> C.meth.__qualname__ |
| 436 | 'C.meth' |
| 437 | |
| 438 | Example with nested classes:: |
| 439 | |
| 440 | >>> class C: |
| 441 | ... class D: |
| 442 | ... def meth(self): |
| 443 | ... pass |
| 444 | ... |
| 445 | >>> C.D.__name__ |
| 446 | 'D' |
| 447 | >>> C.D.__qualname__ |
| 448 | 'C.D' |
| 449 | >>> C.D.meth.__name__ |
| 450 | 'meth' |
| 451 | >>> C.D.meth.__qualname__ |
| 452 | 'C.D.meth' |
| 453 | |
| 454 | Example with nested functions:: |
| 455 | |
| 456 | >>> def outer(): |
| 457 | ... def inner(): |
| 458 | ... pass |
| 459 | ... return inner |
| 460 | ... |
| 461 | >>> outer().__name__ |
| 462 | 'inner' |
| 463 | >>> outer().__qualname__ |
| 464 | 'outer.<locals>.inner' |
| 465 | |
Antoine Pitrou | e7ede06 | 2011-11-25 19:11:26 +0100 | [diff] [blame] | 466 | The string representation of those objects is also changed to include the |
Antoine Pitrou | 6bbd76b | 2011-11-25 19:10:05 +0100 | [diff] [blame] | 467 | new, more precise information:: |
| 468 | |
| 469 | >>> str(C.D) |
| 470 | "<class '__main__.C.D'>" |
| 471 | >>> str(C.D.meth) |
| 472 | '<function C.D.meth at 0x7f46b9fe31e0>' |
| 473 | |
| 474 | |
Brett Cannon | c204348 | 2012-04-29 20:59:41 -0400 | [diff] [blame] | 475 | Using importlib as the Implementation of Import |
| 476 | =============================================== |
| 477 | :issue:`2377` - Replace __import__ w/ importlib.__import__ |
| 478 | :issue:`13959` - Re-implement parts of :mod:`imp` in pure Python |
| 479 | :issue:`14605` - Make import machinery explicit |
| 480 | :issue:`14646` - Require loaders set __loader__ and __package__ |
| 481 | |
| 482 | (Written by Brett Cannon) |
| 483 | |
| 484 | The :func:`__import__` function is now powered by :func:`importlib.__import__`. |
| 485 | This work leads to the completion of "phase 2" of :pep:`302`. There are |
| 486 | multiple benefits to this change. First, it has allowed for more of the |
| 487 | machinery powering import to be exposed instead of being implicit and hidden |
| 488 | within the C code. It also provides a single implementation for all Python VMs |
| 489 | supporting Python 3.3 to use, helping to end any VM-specific deviations in |
| 490 | import semantics. And finally it eases the maintenance of import, allowing for |
| 491 | future growth to occur. |
| 492 | |
| 493 | For the common user, this change should result in no visible change in |
| 494 | semantics. Any possible changes required in one's code to handle this change |
| 495 | should read the `Porting Python code`_ section of this document to see what |
| 496 | needs to be changed, but it will only affect those that currently manipulate |
| 497 | import or try calling it programmatically. |
| 498 | |
| 499 | New APIs |
| 500 | -------- |
| 501 | One of the large benefits of this work is the exposure of what goes into |
| 502 | making the import statement work. That means the various importers that were |
| 503 | once implicit are now fully exposed as part of the :mod:`importlib` package. |
| 504 | |
| 505 | In terms of finders, * :class:`importlib.machinery.FileFinder` exposes the |
| 506 | mechanism used to search for source and bytecode files of a module. Previously |
| 507 | this class was an implicit member of :attr:`sys.path_hooks`. |
| 508 | |
| 509 | For loaders, the new abstract base class :class:`importlib.abc.FileLoader` helps |
| 510 | write a loader that uses the file system as the storage mechanism for a module's |
| 511 | code. The loader for source files |
| 512 | (:class:`importlib.machinery.SourceFileLoader`), sourceless bytecode files |
| 513 | (:class:`importlib.machinery.SourcelessFileLoader`), and extension modules |
| 514 | (:class:`importlib.machinery.ExtensionFileLoader`) are now available for |
| 515 | direct use. |
| 516 | |
| 517 | :exc:`ImportError` now has ``name`` and ``path`` attributes which are set when |
| 518 | there is relevant data to provide. The message for failed imports will also |
| 519 | provide the full name of the module now instead of just the tail end of the |
| 520 | module's name. |
| 521 | |
| 522 | The :func:`importlib.invalidate_caches` function will now call the method with |
| 523 | the same name on all finders cached in :attr:`sys.path_importer_cache` to help |
| 524 | clean up any stored state as necessary. |
| 525 | |
| 526 | Visible Changes |
| 527 | --------------- |
| 528 | [For potential required changes to code, see the `Porting Python code`_ |
| 529 | section] |
| 530 | |
| 531 | Beyond the expanse of what :mod:`importlib` now exposes, there are other |
| 532 | visible changes to import. The biggest is that :attr:`sys.meta_path` and |
| 533 | :attr:`sys.path_hooks` now store all of the finders used by import explicitly. |
| 534 | Previously the finders were implicit and hidden within the C code of import |
| 535 | instead of being directly exposed. This means that one can now easily remove or |
| 536 | change the order of the various finders to fit one's needs. |
| 537 | |
| 538 | Another change is that all modules have a ``__loader__`` attribute, storing the |
| 539 | loader used to create the module. :pep:`302` has been updated to make this |
| 540 | attribute mandatory for loaders to implement, so in the future once 3rd-party |
| 541 | loaders have been updated people will be able to rely on the existence of the |
| 542 | attribute. Until such time, though, import is setting the module post-load. |
| 543 | |
| 544 | Loaders are also now expected to set the ``__package__`` attribute from |
| 545 | :pep:`366`. Once again, import itself is already setting this on all loaders |
| 546 | from :mod:`importlib` and import itself is setting the attribute post-load. |
| 547 | |
| 548 | ``None`` is now inserted into :attr:`sys.path_importer_cache` when no finder |
| 549 | can be found on :attr:`sys.path_hooks`. Since :class:`imp.NullImporter` is not |
| 550 | directly exposed on :attr:`sys.path_hooks` it could no longer be relied upon to |
| 551 | always be available to use as a value representing no finder found. |
| 552 | |
| 553 | All other changes relate to semantic changes which should be taken into |
| 554 | consideration when updating code for Python 3.3, and thus should be read about |
| 555 | in the `Porting Python code`_ section of this document. |
| 556 | |
| 557 | |
R David Murray | 0fa2edd | 2012-05-25 17:59:56 -0400 | [diff] [blame] | 558 | New Email Package Features |
| 559 | ========================== |
| 560 | |
R David Murray | cb448cf | 2012-05-25 22:25:56 -0400 | [diff] [blame] | 561 | Policy Framework |
| 562 | ---------------- |
| 563 | |
R David Murray | 0fa2edd | 2012-05-25 17:59:56 -0400 | [diff] [blame] | 564 | The email package now has a :mod:`~email.policy` framework. A |
| 565 | :class:`~email.policy.Policy` is an object with several methods and properties |
| 566 | that control how the email package behaves. The primary policy for Python 3.3 |
| 567 | is the :class:`~email.policy.Compat32` policy, which provides backward |
| 568 | compatibility with the email package in Python 3.2. A ``policy`` can be |
| 569 | specified when an email message is parsed by a :mod:`~email.parser`, or when a |
| 570 | :class:`~email.message.Message` object is created, or when an email is |
| 571 | serialized using a :mod:`~email.generator`. Unless overridden, a policy passed |
| 572 | to a ``parser`` is inherited by all the ``Message`` object and sub-objects |
| 573 | created by the ``parser``. By default a ``generator`` will use the policy of |
| 574 | the ``Message`` object it is serializing. The default policy is |
| 575 | :data:`~email.policy.compat32`. |
| 576 | |
| 577 | The minimum set of controls implemented by all ``policy`` objects are: |
| 578 | |
| 579 | =============== ======================================================= |
| 580 | max_line_length The maximum length, excluding the linesep character(s), |
| 581 | individual lines may have when a ``Message`` is |
| 582 | serialized. Defaults to 78. |
| 583 | |
| 584 | linesep The character used to separate individual lines when a |
| 585 | ``Message`` is serialized. Defaults to ``\n``. |
| 586 | |
| 587 | cte_type ``7bit`` or ``8bit``. ``8bit`` applies only to a |
| 588 | ``Bytes`` ``generator``, and means that non-ASCII may |
| 589 | be used where allowed by the protocol (or where it |
| 590 | exists in the original input). |
| 591 | |
| 592 | raise_on_defect Causes a ``parser`` to raise error when defects are |
| 593 | encountered instead of adding them to the ``Message`` |
| 594 | object's ``defects`` list. |
| 595 | =============== ======================================================= |
| 596 | |
| 597 | A new policy instance, with new settings, is created using the |
| 598 | :meth:`~email.policy.Policy.clone` method of policy objects. ``clone`` takes |
| 599 | any of the above controls as keyword arguments. Any control not specified in |
| 600 | the call retains its default value. Thus you can create a policy that uses |
| 601 | ``\r\n`` linesep characters like this:: |
| 602 | |
Georg Brandl | 3539afd | 2012-05-30 22:03:20 +0200 | [diff] [blame] | 603 | mypolicy = compat32.clone(linesep='\r\n') |
R David Murray | 0fa2edd | 2012-05-25 17:59:56 -0400 | [diff] [blame] | 604 | |
| 605 | Policies can be used to make the generation of messages in the format needed by |
| 606 | your application simpler. Instead of having to remember to specify |
| 607 | ``linesep='\r\n'`` in all the places you call a ``generator``, you can specify |
| 608 | it once, when you set the policy used by the ``parser`` or the ``Message``, |
| 609 | whichever your program uses to create ``Message`` objects. On the other hand, |
| 610 | if you need to generate messages in multiple forms, you can still specify the |
| 611 | parameters in the appropriate ``generator`` call. Or you can have custom |
| 612 | policy instances for your different cases, and pass those in when you create |
| 613 | the ``generator``. |
| 614 | |
| 615 | |
R David Murray | cb448cf | 2012-05-25 22:25:56 -0400 | [diff] [blame] | 616 | Provisional Policy with New Header API |
| 617 | -------------------------------------- |
| 618 | |
| 619 | While the policy framework is worthwhile all by itself, the main motivation for |
| 620 | introducing it is to allow the creation of new policies that implement new |
| 621 | features for the email package in a way that maintains backward compatibility |
| 622 | for those who do not use the new policies. Because the new policies introduce a |
| 623 | new API, we are releasing them in Python 3.3 as a :term:`provisional policy |
| 624 | <provisional package>`. Backwards incompatible changes (up to and including |
| 625 | removal of the code) may occur if deemed necessary by the core developers. |
| 626 | |
| 627 | The new policies are instances of :class:`~email.policy.EmailPolicy`, |
| 628 | and add the following additional controls: |
| 629 | |
| 630 | =============== ======================================================= |
| 631 | refold_source Controls whether or not headers parsed by a |
| 632 | :mod:`~email.parser` are refolded by the |
| 633 | :mod:`~email.generator`. It can be ``none``, ``long``, |
| 634 | or ``all``. The default is ``long``, which means that |
| 635 | source headers with a line longer than |
| 636 | ``max_line_length`` get refolded. ``none`` means no |
| 637 | line get refolded, and ``all`` means that all lines |
| 638 | get refolded. |
| 639 | |
| 640 | header_factory A callable that take a ``name`` and ``value`` and |
| 641 | produces a custom header object. |
| 642 | =============== ======================================================= |
| 643 | |
| 644 | The ``header_factory`` is the key to the new features provided by the new |
| 645 | policies. When one of the new policies is used, any header retrieved from |
| 646 | a ``Message`` object is an object produced by the ``header_factory``, and any |
| 647 | time you set a header on a ``Message`` it becomes an object produced by |
| 648 | ``header_factory``. All such header objects have a ``name`` attribute equal |
| 649 | to the header name. Address and Date headers have additional attributes |
| 650 | that give you access to the parsed data of the header. This means you can now |
| 651 | do things like this:: |
| 652 | |
| 653 | >>> m = Message(policy=SMTP) |
| 654 | >>> m['To'] = 'Éric <foo@example.com>' |
| 655 | >>> m['to'] |
| 656 | 'Éric <foo@example.com>' |
| 657 | >>> m['to'].addresses |
| 658 | (Address(display_name='Éric', username='foo', domain='example.com'),) |
| 659 | >>> m['to'].addresses[0].username |
| 660 | 'foo' |
| 661 | >>> m['to'].addresses[0].display_name |
| 662 | 'Éric' |
| 663 | >>> m['Date'] = email.utils.localtime() |
| 664 | >>> m['Date'].datetime |
| 665 | datetime.datetime(2012, 5, 25, 21, 39, 24, 465484, tzinfo=datetime.timezone(datetime.timedelta(-1, 72000), 'EDT')) |
| 666 | >>> m['Date'] |
| 667 | 'Fri, 25 May 2012 21:44:27 -0400' |
| 668 | >>> print(m) |
| 669 | To: =?utf-8?q?=C3=89ric?= <foo@example.com> |
| 670 | Date: Fri, 25 May 2012 21:44:27 -0400 |
| 671 | |
| 672 | You will note that the unicode display name is automatically encoded as |
| 673 | ``utf-8`` when the message is serialized, but that when the header is accessed |
| 674 | directly, you get the unicode version. This eliminates any need to deal with |
| 675 | the :mod:`email.header` :meth:`~email.header.decode_header` or |
| 676 | :meth:`~email.header.make_header` functions. |
| 677 | |
| 678 | You can also create addresses from parts:: |
| 679 | |
| 680 | >>> m['cc'] = [Group('pals', [Address('Bob', 'bob', 'example.com'), |
| 681 | ... Address('Sally', 'sally', 'example.com')]), |
| 682 | ... Address('Bonzo', addr_spec='bonz@laugh.com')] |
| 683 | >>> print(m) |
| 684 | To: =?utf-8?q?=C3=89ric?= <foo@example.com> |
| 685 | Date: Fri, 25 May 2012 21:44:27 -0400 |
| 686 | cc: pals: Bob <bob@example.com>, Sally <sally@example.com>;, Bonzo <bonz@laugh.com> |
| 687 | |
| 688 | Decoding to unicode is done automatically:: |
| 689 | |
| 690 | >>> m2 = message_from_string(str(m)) |
| 691 | >>> m2['to'] |
| 692 | 'Éric <foo@example.com>' |
| 693 | |
| 694 | When you parse a message, you can use the ``addresses`` and ``groups`` |
| 695 | attributes of the header objects to access the groups and individual |
| 696 | addresses:: |
| 697 | |
| 698 | >>> m2['cc'].addresses |
| 699 | (Address(display_name='Bob', username='bob', domain='example.com'), Address(display_name='Sally', username='sally', domain='example.com'), Address(display_name='Bonzo', username='bonz', domain='laugh.com')) |
| 700 | >>> m2['cc'].groups |
| 701 | (Group(display_name='pals', addresses=(Address(display_name='Bob', username='bob', domain='example.com'), Address(display_name='Sally', username='sally', domain='example.com')), Group(display_name=None, addresses=(Address(display_name='Bonzo', username='bonz', domain='laugh.com'),)) |
| 702 | |
| 703 | In summary, if you use one of the new policies, header manipulation works the |
| 704 | way it ought to: your application works with unicode strings, and the email |
| 705 | package transparently encodes and decodes the unicode to and from the RFC |
| 706 | standard Content Transfer Encodings. |
| 707 | |
| 708 | |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 709 | Other Language Changes |
| 710 | ====================== |
| 711 | |
| 712 | Some smaller changes made to the core Python language are: |
| 713 | |
Antoine Pitrou | 7b578b3 | 2011-11-29 22:47:11 +0100 | [diff] [blame] | 714 | * Added support for Unicode name aliases and named sequences. |
| 715 | Both :func:`unicodedata.lookup()` and ``'\N{...}'`` now resolve name aliases, |
| 716 | and :func:`unicodedata.lookup()` resolves named sequences too. |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 717 | |
Antoine Pitrou | 7b578b3 | 2011-11-29 22:47:11 +0100 | [diff] [blame] | 718 | (Contributed by Ezio Melotti in :issue:`12753`) |
Ezio Melotti | 931b8aa | 2011-10-21 21:57:36 +0300 | [diff] [blame] | 719 | |
Antoine Pitrou | 7b578b3 | 2011-11-29 22:47:11 +0100 | [diff] [blame] | 720 | * Equality comparisons on :func:`range` objects now return a result reflecting |
| 721 | the equality of the underlying sequences generated by those range objects. |
Ezio Melotti | 931b8aa | 2011-10-21 21:57:36 +0300 | [diff] [blame] | 722 | |
Sandro Tosi | cd89912 | 2012-01-22 12:16:04 +0100 | [diff] [blame] | 723 | (:issue:`13201`) |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 724 | |
Antoine Pitrou | 7b578b3 | 2011-11-29 22:47:11 +0100 | [diff] [blame] | 725 | * The ``count()``, ``find()``, ``rfind()``, ``index()`` and ``rindex()`` |
| 726 | methods of :class:`bytes` and :class:`bytearray` objects now accept an |
| 727 | integer between 0 and 255 as their first argument. |
Mark Dickinson | 3664568 | 2011-10-23 19:53:01 +0100 | [diff] [blame] | 728 | |
Antoine Pitrou | 7b578b3 | 2011-11-29 22:47:11 +0100 | [diff] [blame] | 729 | (:issue:`12170`) |
Mark Dickinson | 3664568 | 2011-10-23 19:53:01 +0100 | [diff] [blame] | 730 | |
Eli Bendersky | 7add4ea | 2012-03-17 15:14:35 +0200 | [diff] [blame] | 731 | * New methods have been added to :class:`list` and :class:`bytearray`: |
| 732 | ``copy()`` and ``clear()``. |
| 733 | |
| 734 | (:issue:`10516`) |
Petri Lehtinen | 61ea8a0 | 2011-11-24 22:00:46 +0200 | [diff] [blame] | 735 | |
Antoine Pitrou | 9a86447 | 2012-05-04 23:15:47 +0200 | [diff] [blame] | 736 | * Raw bytes literals can now be written ``rb"..."`` as well as ``br"..."``. |
| 737 | (Contributed by Antoine Pitrou in :issue:`13748`.) |
| 738 | |
| 739 | * :meth:`dict.setdefault` now does only one lookup for the given key, making |
| 740 | it atomic when used with built-in types. |
| 741 | (Contributed by Filip Gruszczyński in :issue:`13521`.) |
| 742 | |
| 743 | |
Benjamin Peterson | e50d6ab | 2012-04-03 00:52:18 -0400 | [diff] [blame] | 744 | .. XXX mention new error messages for passing wrong number of arguments to functions |
| 745 | |
Antoine Pitrou | 9a86447 | 2012-05-04 23:15:47 +0200 | [diff] [blame] | 746 | |
Antoine Pitrou | 79341e7 | 2012-05-17 21:13:45 +0200 | [diff] [blame] | 747 | A Finer-Grained Import Lock |
| 748 | =========================== |
| 749 | |
| 750 | Previous versions of CPython have always relied on a global import lock. |
| 751 | This led to unexpected annoyances, such as deadlocks when importing a module |
| 752 | would trigger code execution in a different thread as a side-effect. |
| 753 | Clumsy workarounds were sometimes employed, such as the |
| 754 | :c:func:`PyImport_ImportModuleNoBlock` C API function. |
| 755 | |
| 756 | In Python 3.3, importing a module takes a per-module lock. This correctly |
| 757 | serializes importation of a given module from multiple threads (preventing |
| 758 | the exposure of incompletely initialized modules), while eliminating the |
| 759 | aforementioned annoyances. |
| 760 | |
| 761 | (contributed by Antoine Pitrou in :issue:`9260`.) |
| 762 | |
| 763 | |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 764 | New and Improved Modules |
| 765 | ======================== |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 766 | |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 767 | abc |
| 768 | --- |
| 769 | |
| 770 | Improved support for abstract base classes containing descriptors composed with |
| 771 | abstract methods. The recommended approach to declaring abstract descriptors is |
| 772 | now to provide :attr:`__isabstractmethod__` as a dynamically updated |
| 773 | property. The built-in descriptors have been updated accordingly. |
| 774 | |
| 775 | * :class:`abc.abstractproperty` has been deprecated, use :class:`property` |
| 776 | with :func:`abc.abstractmethod` instead. |
| 777 | * :class:`abc.abstractclassmethod` has been deprecated, use |
| 778 | :class:`classmethod` with :func:`abc.abstractmethod` instead. |
| 779 | * :class:`abc.abstractstaticmethod` has been deprecated, use |
| 780 | :class:`staticmethod` with :func:`abc.abstractmethod` instead. |
| 781 | |
| 782 | (Contributed by Darren Dale in :issue:`11610`) |
| 783 | |
Meador Inge | c5dbb3d | 2011-09-20 21:48:16 -0500 | [diff] [blame] | 784 | array |
| 785 | ----- |
| 786 | |
| 787 | The :mod:`array` module supports the :c:type:`long long` type using ``q`` and |
| 788 | ``Q`` type codes. |
| 789 | |
| 790 | (Contributed by Oren Tirosh and Hirokazu Yamamoto in :issue:`1172711`) |
| 791 | |
| 792 | |
Nadeem Vawda | d7e5c6e | 2012-02-12 01:34:18 +0200 | [diff] [blame] | 793 | bz2 |
| 794 | --- |
| 795 | |
| 796 | The :mod:`bz2` module has been rewritten from scratch. In the process, several |
| 797 | new features have been added: |
| 798 | |
| 799 | * :class:`bz2.BZ2File` can now read from and write to arbitrary file-like |
| 800 | objects, by means of its constructor's *fileobj* argument. |
| 801 | |
| 802 | (Contributed by Nadeem Vawda in :issue:`5863`) |
| 803 | |
| 804 | * :class:`bz2.BZ2File` and :func:`bz2.decompress` can now decompress |
| 805 | multi-stream inputs (such as those produced by the :program:`pbzip2` tool). |
| 806 | :class:`bz2.BZ2File` can now also be used to create this type of file, using |
| 807 | the ``'a'`` (append) mode. |
| 808 | |
| 809 | (Contributed by Nir Aides in :issue:`1625`) |
| 810 | |
| 811 | * :class:`bz2.BZ2File` now implements all of the :class:`io.BufferedIOBase` API, |
| 812 | except for the :meth:`detach` and :meth:`truncate` methods. |
| 813 | |
| 814 | |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 815 | codecs |
| 816 | ------ |
| 817 | |
Antoine Pitrou | 4f86343 | 2012-02-12 02:12:47 +0100 | [diff] [blame] | 818 | The :mod:`~encodings.mbcs` codec has been rewritten to handle correctly |
Georg Brandl | ff962c5 | 2012-02-04 08:55:56 +0100 | [diff] [blame] | 819 | ``replace`` and ``ignore`` error handlers on all Windows versions. The |
| 820 | :mod:`~encodings.mbcs` codec now supports all error handlers, instead of only |
| 821 | ``replace`` to encode and ``ignore`` to decode. |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 822 | |
Georg Brandl | ff962c5 | 2012-02-04 08:55:56 +0100 | [diff] [blame] | 823 | A new Windows-only codec has been added: ``cp65001`` (:issue:`13216`). It is the |
| 824 | Windows code page 65001 (Windows UTF-8, ``CP_UTF8``). For example, it is used |
| 825 | by ``sys.stdout`` if the console output code page is set to cp65001 (e.g., using |
| 826 | ``chcp 65001`` command). |
Victor Stinner | 2f3ca9f | 2011-10-27 01:38:56 +0200 | [diff] [blame] | 827 | |
Georg Brandl | ff962c5 | 2012-02-04 08:55:56 +0100 | [diff] [blame] | 828 | Multibyte CJK decoders now resynchronize faster. They only ignore the first |
Georg Brandl | 6c0929b | 2011-07-09 11:43:33 +0200 | [diff] [blame] | 829 | byte of an invalid byte sequence. For example, ``b'\xff\n'.decode('gb2312', |
| 830 | 'replace')`` now returns a ``\n`` after the replacement character. |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 831 | |
Georg Brandl | 6c0929b | 2011-07-09 11:43:33 +0200 | [diff] [blame] | 832 | (:issue:`12016`) |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 833 | |
Georg Brandl | ff962c5 | 2012-02-04 08:55:56 +0100 | [diff] [blame] | 834 | Incremental CJK codec encoders are no longer reset at each call to their |
| 835 | encode() methods. For example:: |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 836 | |
| 837 | $ ./python -q |
| 838 | >>> import codecs |
| 839 | >>> encoder = codecs.getincrementalencoder('hz')('strict') |
| 840 | >>> b''.join(encoder.encode(x) for x in '\u52ff\u65bd\u65bc\u4eba\u3002 Bye.') |
| 841 | b'~{NpJ)l6HK!#~} Bye.' |
| 842 | |
Georg Brandl | 6c0929b | 2011-07-09 11:43:33 +0200 | [diff] [blame] | 843 | This example gives ``b'~{Np~}~{J)~}~{l6~}~{HK~}~{!#~} Bye.'`` with older Python |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 844 | versions. |
| 845 | |
Georg Brandl | 6c0929b | 2011-07-09 11:43:33 +0200 | [diff] [blame] | 846 | (:issue:`12100`) |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 847 | |
Victor Stinner | 9f4b1e9 | 2011-11-10 20:56:30 +0100 | [diff] [blame] | 848 | The ``unicode_internal`` codec has been deprecated. |
| 849 | |
Éric Araujo | 4f61a2d | 2012-04-04 23:01:01 -0400 | [diff] [blame] | 850 | |
| 851 | collections |
| 852 | ----------- |
| 853 | |
| 854 | Addition of a new :class:`~collections.ChainMap` class to allow treating a |
| 855 | number of mappings as a single unit. |
| 856 | |
| 857 | (Written by Raymond Hettinger for :issue:`11089`, made public in |
| 858 | :issue:`11297`) |
| 859 | |
| 860 | The abstract base classes have been moved in a new :mod:`collections.abc` |
| 861 | module, to better differentiate between the abstract and the concrete |
| 862 | collections classes. Aliases for ABCs are still present in the |
| 863 | :mod:`collections` module to preserve existing imports. |
| 864 | |
| 865 | (:issue:`11085`) |
| 866 | |
| 867 | .. XXX addition of __slots__ to ABCs not recorded here: internal detail |
| 868 | |
| 869 | |
Nick Coghlan | 3267a30 | 2012-05-21 22:54:43 +1000 | [diff] [blame] | 870 | contextlib |
| 871 | ---------- |
| 872 | |
| 873 | :class:`~collections.ExitStack` now provides a solid foundation for |
| 874 | programmatic manipulation of context managers and similar cleanup |
| 875 | functionality. Unlike the previous ``contextlib.nested`` API (which was |
| 876 | deprecated and removed), the new API is designed to work correctly |
| 877 | regardless of whether context managers acquire their resources in |
Nick Coghlan | 161ea6a | 2012-05-22 23:04:42 +1000 | [diff] [blame] | 878 | their ``__init__`` method (for example, file objects) or in their |
Nick Coghlan | 3267a30 | 2012-05-21 22:54:43 +1000 | [diff] [blame] | 879 | ``__enter__`` method (for example, synchronisation objects from the |
| 880 | :mod:`threading` module). |
| 881 | |
| 882 | (:issue:`13585`) |
| 883 | |
| 884 | |
Éric Araujo | 84b8ed8 | 2011-08-29 21:42:47 +0200 | [diff] [blame] | 885 | crypt |
| 886 | ----- |
| 887 | |
Victor Stinner | c78fb33 | 2011-09-21 03:35:44 +0200 | [diff] [blame] | 888 | Addition of salt and modular crypt format and the :func:`~crypt.mksalt` |
| 889 | function to the :mod:`crypt` module. |
Éric Araujo | 84b8ed8 | 2011-08-29 21:42:47 +0200 | [diff] [blame] | 890 | |
| 891 | (:issue:`10924`) |
| 892 | |
Victor Stinner | a7878b7 | 2011-07-14 23:07:44 +0200 | [diff] [blame] | 893 | curses |
| 894 | ------ |
| 895 | |
Victor Stinner | 0fdfceb | 2011-11-25 22:10:02 +0100 | [diff] [blame] | 896 | * If the :mod:`curses` module is linked to the ncursesw library, use Unicode |
| 897 | functions when Unicode strings or characters are passed (e.g. |
| 898 | :c:func:`waddwstr`), and bytes functions otherwise (e.g. :c:func:`waddstr`). |
| 899 | * Use the locale encoding instead of ``utf-8`` to encode Unicode strings. |
| 900 | * :class:`curses.window` has a new :attr:`curses.window.encoding` attribute. |
Victor Stinner | c78fb33 | 2011-09-21 03:35:44 +0200 | [diff] [blame] | 901 | * The :class:`curses.window` class has a new :meth:`~curses.window.get_wch` |
| 902 | method to get a wide character |
| 903 | * The :mod:`curses` module has a new :meth:`~curses.unget_wch` function to |
| 904 | push a wide character so the next :meth:`~curses.window.get_wch` will return |
| 905 | it |
Victor Stinner | a7878b7 | 2011-07-14 23:07:44 +0200 | [diff] [blame] | 906 | |
Victor Stinner | c78fb33 | 2011-09-21 03:35:44 +0200 | [diff] [blame] | 907 | (Contributed by Iñigo Serna in :issue:`6755`) |
Victor Stinner | a7878b7 | 2011-07-14 23:07:44 +0200 | [diff] [blame] | 908 | |
Stefan Krah | 1919b7e | 2012-03-21 18:25:23 +0100 | [diff] [blame] | 909 | decimal |
| 910 | ------- |
| 911 | |
| 912 | :issue:`7652` - integrate fast native decimal arithmetic. |
| 913 | C-module and libmpdec written by Stefan Krah. |
| 914 | |
| 915 | The new C version of the decimal module integrates the high speed libmpdec |
Stefan Krah | bf80308 | 2012-04-01 13:07:24 +0200 | [diff] [blame] | 916 | library for arbitrary precision correctly-rounded decimal floating point |
| 917 | arithmetic. libmpdec conforms to IBM's General Decimal Arithmetic Specification. |
Stefan Krah | 1919b7e | 2012-03-21 18:25:23 +0100 | [diff] [blame] | 918 | |
Stefan Krah | 0c0914e | 2012-04-09 20:31:15 +0200 | [diff] [blame] | 919 | Performance gains range from 10x for database applications to 100x for |
Stefan Krah | bf80308 | 2012-04-01 13:07:24 +0200 | [diff] [blame] | 920 | numerically intensive applications. These numbers are expected gains |
| 921 | for standard precisions used in decimal floating point arithmetic. Since |
| 922 | the precision is user configurable, the exact figures may vary. For example, |
| 923 | in integer bignum arithmetic the differences can be significantly higher. |
| 924 | |
| 925 | The following table is meant as an illustration. Benchmarks are available |
Georg Brandl | 204e789 | 2012-04-01 13:10:58 +0200 | [diff] [blame] | 926 | at http://www.bytereef.org/mpdecimal/quickstart.html. |
Stefan Krah | 1919b7e | 2012-03-21 18:25:23 +0100 | [diff] [blame] | 927 | |
| 928 | +---------+-------------+--------------+-------------+ |
| 929 | | | decimal.py | _decimal | speedup | |
| 930 | +=========+=============+==============+=============+ |
Stefan Krah | 0c0914e | 2012-04-09 20:31:15 +0200 | [diff] [blame] | 931 | | pi | 38.89s | 0.38s | 100x | |
Stefan Krah | 1919b7e | 2012-03-21 18:25:23 +0100 | [diff] [blame] | 932 | +---------+-------------+--------------+-------------+ |
| 933 | | telco | 172.19s | 5.68s | 30x | |
| 934 | +---------+-------------+--------------+-------------+ |
| 935 | | psycopg | 3.57s | 0.29s | 12x | |
| 936 | +---------+-------------+--------------+-------------+ |
| 937 | |
| 938 | Features |
| 939 | ~~~~~~~~ |
| 940 | |
| 941 | * The :exc:`~decimal.FloatOperation` signal optionally enables stricter |
| 942 | semantics for mixing floats and Decimals. |
| 943 | |
| 944 | * If Python is compiled without threads, the C version automatically |
| 945 | disables the expensive thread local context machinery. In this case, |
| 946 | the variable :data:`~decimal.HAVE_THREADS` is set to False. |
| 947 | |
| 948 | API changes |
| 949 | ~~~~~~~~~~~ |
| 950 | |
| 951 | * The C module has the following context limits, depending on the machine |
| 952 | architecture: |
| 953 | |
| 954 | +-------------------+---------------------+------------------------------+ |
| 955 | | | 32-bit | 64-bit | |
| 956 | +===================+=====================+==============================+ |
| 957 | | :const:`MAX_PREC` | :const:`425000000` | :const:`999999999999999999` | |
| 958 | +-------------------+---------------------+------------------------------+ |
| 959 | | :const:`MAX_EMAX` | :const:`425000000` | :const:`999999999999999999` | |
| 960 | +-------------------+---------------------+------------------------------+ |
| 961 | | :const:`MIN_EMIN` | :const:`-425000000` | :const:`-999999999999999999` | |
| 962 | +-------------------+---------------------+------------------------------+ |
| 963 | |
| 964 | * In the context templates (:class:`~decimal.DefaultContext`, |
| 965 | :class:`~decimal.BasicContext` and :class:`~decimal.ExtendedContext`) |
| 966 | the magnitude of :attr:`~decimal.Context.Emax` and |
| 967 | :attr:`~decimal.Context.Emin` has changed to :const:`999999`. |
| 968 | |
| 969 | * The :class:`~decimal.Decimal` constructor in decimal.py does not observe |
| 970 | the context limits and converts values with arbitrary exponents or precision |
| 971 | exactly. Since the C version has internal limits, the following scheme is |
| 972 | used: If possible, values are converted exactly, otherwise |
| 973 | :exc:`~decimal.InvalidOperation` is raised and the result is NaN. In the |
| 974 | latter case it is always possible to use :meth:`~decimal.Context.create_decimal` |
| 975 | in order to obtain a rounded or inexact value. |
| 976 | |
| 977 | |
| 978 | * The power function in decimal.py is always correctly-rounded. In the |
| 979 | C version, it is defined in terms of the correctly-rounded |
| 980 | :meth:`~decimal.Decimal.exp` and :meth:`~decimal.Decimal.ln` functions, |
| 981 | but the final result is only "almost always correctly rounded". |
| 982 | |
| 983 | |
| 984 | * In the C version, the context dictionary containing the signals is a |
| 985 | :class:`~collections.abc.MutableMapping`. For speed reasons, |
| 986 | :attr:`~decimal.Context.flags` and :attr:`~decimal.Context.traps` always |
| 987 | refer to the same :class:`~collections.abc.MutableMapping` that the context |
| 988 | was initialized with. If a new signal dictionary is assigned, |
| 989 | :attr:`~decimal.Context.flags` and :attr:`~decimal.Context.traps` |
| 990 | are updated with the new values, but they do not reference the RHS |
| 991 | dictionary. |
| 992 | |
| 993 | |
| 994 | * Pickling a :class:`~decimal.Context` produces a different output in order |
| 995 | to have a common interchange format for the Python and C versions. |
| 996 | |
| 997 | |
| 998 | * The order of arguments in the :class:`~decimal.Context` constructor has been |
| 999 | changed to match the order displayed by :func:`repr`. |
| 1000 | |
| 1001 | |
Victor Stinner | 024e37a | 2011-03-31 01:31:06 +0200 | [diff] [blame] | 1002 | faulthandler |
| 1003 | ------------ |
| 1004 | |
| 1005 | New module: :mod:`faulthandler`. |
| 1006 | |
| 1007 | * :envvar:`PYTHONFAULTHANDLER` |
| 1008 | * :option:`-X` ``faulthandler`` |
| 1009 | |
Victor Stinner | 811db3b | 2011-09-21 03:20:03 +0200 | [diff] [blame] | 1010 | ftplib |
| 1011 | ------ |
| 1012 | |
| 1013 | The :class:`~ftplib.FTP_TLS` class now provides a new |
| 1014 | :func:`~ftplib.FTP_TLS.ccc` function to revert control channel back to |
Florent Xicluna | 6d57d21 | 2011-10-23 22:23:57 +0200 | [diff] [blame] | 1015 | plaintext. This can be useful to take advantage of firewalls that know how to |
Victor Stinner | 811db3b | 2011-09-21 03:20:03 +0200 | [diff] [blame] | 1016 | handle NAT with non-secure FTP without opening fixed ports. |
| 1017 | |
| 1018 | (Contributed by Giampaolo Rodolà in :issue:`12139`) |
| 1019 | |
| 1020 | |
Antoine Pitrou | 5a8bc6f | 2011-11-17 02:20:48 +0100 | [diff] [blame] | 1021 | imaplib |
| 1022 | ------- |
| 1023 | |
| 1024 | The :class:`~imaplib.IMAP4_SSL` constructor now accepts an SSLContext |
| 1025 | parameter to control parameters of the secure channel. |
| 1026 | |
| 1027 | (Contributed by Sijin Joseph in :issue:`8808`) |
| 1028 | |
| 1029 | |
Charles-François Natali | dc3044c | 2012-01-09 22:40:02 +0100 | [diff] [blame] | 1030 | io |
| 1031 | -- |
| 1032 | |
Charles-François Natali | d612de1 | 2012-01-14 11:51:00 +0100 | [diff] [blame] | 1033 | The :func:`~io.open` function has a new ``'x'`` mode that can be used to |
| 1034 | exclusively create a new file, and raise a :exc:`FileExistsError` if the file |
| 1035 | already exists. It is based on the C11 'x' mode to fopen(). |
Charles-François Natali | dc3044c | 2012-01-09 22:40:02 +0100 | [diff] [blame] | 1036 | |
| 1037 | (Contributed by David Townshend in :issue:`12760`) |
| 1038 | |
| 1039 | |
Nick Coghlan | dc9b255 | 2012-05-20 21:01:57 +1000 | [diff] [blame] | 1040 | ipaddress |
| 1041 | --------- |
| 1042 | |
| 1043 | The new :mod:`ipaddress` module provides tools for creating and manipulating |
| 1044 | objects representing IPv4 and IPv6 addresses, networks and interfaces (i.e. |
| 1045 | an IP address associated with a specific IP subnet). |
| 1046 | |
| 1047 | (Contributed by Google and Peter Moody in :pep:`3144`) |
| 1048 | |
Nadeem Vawda | 3459922 | 2011-12-09 01:32:46 +0200 | [diff] [blame] | 1049 | lzma |
| 1050 | ---- |
| 1051 | |
| 1052 | The newly-added :mod:`lzma` module provides data compression and decompression |
| 1053 | using the LZMA algorithm, including support for the ``.xz`` and ``.lzma`` |
| 1054 | file formats. |
| 1055 | |
| 1056 | (Contributed by Nadeem Vawda and Per Øyvind Karlsen in :issue:`6715`) |
| 1057 | |
| 1058 | |
Victor Stinner | fa0e3d5 | 2011-05-09 01:01:09 +0200 | [diff] [blame] | 1059 | math |
| 1060 | ---- |
| 1061 | |
| 1062 | The :mod:`math` module has a new function: |
| 1063 | |
| 1064 | * :func:`~math.log2`: return the base-2 logarithm of *x* |
| 1065 | (Written by Mark Dickinson in :issue:`11888`). |
| 1066 | |
| 1067 | |
Antoine Pitrou | 9a86447 | 2012-05-04 23:15:47 +0200 | [diff] [blame] | 1068 | multiprocessing |
| 1069 | --------------- |
| 1070 | |
| 1071 | The new :func:`multiprocessing.connection.wait` function allows to poll |
| 1072 | multiple objects (such as connections, sockets and pipes) with a timeout. |
| 1073 | (Contributed by Richard Oudkerk in :issue:`12328`.) |
| 1074 | |
| 1075 | :class:`multiprocessing.Connection` objects can now be transferred over |
| 1076 | multiprocessing connections. |
| 1077 | (Contributed by Richard Oudkerk in :issue:`4892`.) |
| 1078 | |
| 1079 | |
Victor Stinner | fa0e3d5 | 2011-05-09 01:01:09 +0200 | [diff] [blame] | 1080 | nntplib |
| 1081 | ------- |
| 1082 | |
| 1083 | The :class:`nntplib.NNTP` class now supports the context manager protocol to |
| 1084 | unconditionally consume :exc:`socket.error` exceptions and to close the NNTP |
| 1085 | connection when done:: |
| 1086 | |
| 1087 | >>> from nntplib import NNTP |
Ezio Melotti | 3c14b4e | 2011-07-13 11:44:44 +0300 | [diff] [blame] | 1088 | >>> with NNTP('news.gmane.org') as n: |
Victor Stinner | fa0e3d5 | 2011-05-09 01:01:09 +0200 | [diff] [blame] | 1089 | ... n.group('gmane.comp.python.committers') |
| 1090 | ... |
Ezio Melotti | 04f648c | 2011-07-26 09:37:46 +0300 | [diff] [blame] | 1091 | ('211 1755 1 1755 gmane.comp.python.committers', 1755, 1, 1755, 'gmane.comp.python.committers') |
Victor Stinner | fa0e3d5 | 2011-05-09 01:01:09 +0200 | [diff] [blame] | 1092 | >>> |
| 1093 | |
| 1094 | (Contributed by Giampaolo Rodolà in :issue:`9795`) |
| 1095 | |
| 1096 | |
Giampaolo Rodolà | c9c2c8b | 2011-02-25 14:39:16 +0000 | [diff] [blame] | 1097 | os |
| 1098 | -- |
| 1099 | |
Charles-François Natali | a003af1 | 2011-06-01 20:30:52 +0200 | [diff] [blame] | 1100 | * The :mod:`os` module has a new :func:`~os.pipe2` function that makes it |
| 1101 | possible to create a pipe with :data:`~os.O_CLOEXEC` or |
| 1102 | :data:`~os.O_NONBLOCK` flags set atomically. This is especially useful to |
| 1103 | avoid race conditions in multi-threaded programs. |
| 1104 | |
Giampaolo Rodolà | 18e8bcb | 2011-02-25 20:57:54 +0000 | [diff] [blame] | 1105 | * The :mod:`os` module has a new :func:`~os.sendfile` function which provides |
| 1106 | an efficent "zero-copy" way for copying data from one file (or socket) |
| 1107 | descriptor to another. The phrase "zero-copy" refers to the fact that all of |
| 1108 | the copying of data between the two descriptors is done entirely by the |
| 1109 | kernel, with no copying of data into userspace buffers. :func:`~os.sendfile` |
| 1110 | can be used to efficiently copy data from a file on disk to a network socket, |
| 1111 | e.g. for downloading a file. |
Giampaolo Rodolà | c9c2c8b | 2011-02-25 14:39:16 +0000 | [diff] [blame] | 1112 | |
Giampaolo Rodolà | 18e8bcb | 2011-02-25 20:57:54 +0000 | [diff] [blame] | 1113 | (Patch submitted by Ross Lagerwall and Giampaolo Rodolà in :issue:`10882`.) |
| 1114 | |
| 1115 | * The :mod:`os` module has two new functions: :func:`~os.getpriority` and |
| 1116 | :func:`~os.setpriority`. They can be used to get or set process |
| 1117 | niceness/priority in a fashion similar to :func:`os.nice` but extended to all |
| 1118 | processes instead of just the current one. |
| 1119 | |
| 1120 | (Patch submitted by Giampaolo Rodolà in :issue:`10784`.) |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 1121 | |
Charles-François Natali | 7372b06 | 2012-02-05 15:15:38 +0100 | [diff] [blame] | 1122 | * The :mod:`os` module has a new :func:`~os.fwalk` function similar to |
| 1123 | :func:`~os.walk` except that it also yields file descriptors referring to the |
| 1124 | directories visited. This is especially useful to avoid symlink races. |
| 1125 | |
Antoine Pitrou | 9a86447 | 2012-05-04 23:15:47 +0200 | [diff] [blame] | 1126 | * The new :func:`os.replace` function allows cross-platform renaming of a |
| 1127 | file with overwriting the destination. With :func:`os.rename`, an existing |
| 1128 | destination file is overwritten under POSIX, but raises an error under |
| 1129 | Windows. |
| 1130 | (Contributed by Antoine Pitrou in :issue:`8828`.) |
| 1131 | |
| 1132 | * The new :func:`os.get_terminal_size` function queries the size of the |
| 1133 | terminal attached to a file descriptor. |
| 1134 | (Contributed by Zbigniew Jędrzejewski-Szmek in :issue:`13609`.) |
| 1135 | |
Victor Stinner | e506437 | 2011-10-14 00:08:29 +0200 | [diff] [blame] | 1136 | * "at" functions (:issue:`4761`): |
| 1137 | |
| 1138 | * :func:`~os.faccessat` |
| 1139 | * :func:`~os.fchmodat` |
| 1140 | * :func:`~os.fchownat` |
| 1141 | * :func:`~os.fstatat` |
| 1142 | * :func:`~os.futimesat` |
Victor Stinner | e506437 | 2011-10-14 00:08:29 +0200 | [diff] [blame] | 1143 | * :func:`~os.linkat` |
| 1144 | * :func:`~os.mkdirat` |
| 1145 | * :func:`~os.mkfifoat` |
| 1146 | * :func:`~os.mknodat` |
| 1147 | * :func:`~os.openat` |
| 1148 | * :func:`~os.readlinkat` |
| 1149 | * :func:`~os.renameat` |
| 1150 | * :func:`~os.symlinkat` |
| 1151 | * :func:`~os.unlinkat` |
| 1152 | * :func:`~os.utimensat` |
Victor Stinner | e506437 | 2011-10-14 00:08:29 +0200 | [diff] [blame] | 1153 | |
| 1154 | * extended attributes (:issue:`12720`): |
| 1155 | |
| 1156 | * :func:`~os.fgetxattr` |
| 1157 | * :func:`~os.flistxattr` |
| 1158 | * :func:`~os.fremovexattr` |
| 1159 | * :func:`~os.fsetxattr` |
| 1160 | * :func:`~os.getxattr` |
| 1161 | * :func:`~os.lgetxattr` |
| 1162 | * :func:`~os.listxattr` |
| 1163 | * :func:`~os.llistxattr` |
| 1164 | * :func:`~os.lremovexattr` |
| 1165 | * :func:`~os.lsetxattr` |
| 1166 | * :func:`~os.removexattr` |
| 1167 | * :func:`~os.setxattr` |
| 1168 | |
| 1169 | * Scheduler functions (:issue:`12655`): |
| 1170 | |
| 1171 | * :func:`~os.sched_get_priority_max` |
| 1172 | * :func:`~os.sched_get_priority_min` |
| 1173 | * :func:`~os.sched_getaffinity` |
| 1174 | * :func:`~os.sched_getparam` |
| 1175 | * :func:`~os.sched_getscheduler` |
| 1176 | * :func:`~os.sched_rr_get_interval` |
| 1177 | * :func:`~os.sched_setaffinity` |
| 1178 | * :func:`~os.sched_setparam` |
| 1179 | * :func:`~os.sched_setscheduler` |
| 1180 | * :func:`~os.sched_yield` |
| 1181 | |
| 1182 | * Add some extra posix functions to the os module (:issue:`10812`): |
| 1183 | |
| 1184 | * :func:`~os.fexecve` |
| 1185 | * :func:`~os.futimens` |
Victor Stinner | e506437 | 2011-10-14 00:08:29 +0200 | [diff] [blame] | 1186 | * :func:`~os.futimes` |
| 1187 | * :func:`~os.lockf` |
| 1188 | * :func:`~os.lutimes` |
Victor Stinner | e506437 | 2011-10-14 00:08:29 +0200 | [diff] [blame] | 1189 | * :func:`~os.posix_fadvise` |
| 1190 | * :func:`~os.posix_fallocate` |
| 1191 | * :func:`~os.pread` |
| 1192 | * :func:`~os.pwrite` |
| 1193 | * :func:`~os.readv` |
| 1194 | * :func:`~os.sync` |
| 1195 | * :func:`~os.truncate` |
| 1196 | * :func:`~os.waitid` |
| 1197 | * :func:`~os.writev` |
| 1198 | |
| 1199 | * Other new functions: |
| 1200 | |
Charles-François Natali | 7794090 | 2012-02-06 19:54:48 +0100 | [diff] [blame] | 1201 | * :func:`~os.flistdir` (:issue:`10755`) |
Victor Stinner | e506437 | 2011-10-14 00:08:29 +0200 | [diff] [blame] | 1202 | * :func:`~os.getgrouplist` (:issue:`9344`) |
| 1203 | |
Giampaolo Rodolà | 424298a | 2011-03-03 18:34:06 +0000 | [diff] [blame] | 1204 | |
Éric Araujo | 765e94f | 2011-06-03 17:26:59 +0200 | [diff] [blame] | 1205 | packaging |
| 1206 | --------- |
| 1207 | |
| 1208 | :mod:`distutils` has undergone additions and refactoring under a new name, |
Éric Araujo | 4f61a2d | 2012-04-04 23:01:01 -0400 | [diff] [blame] | 1209 | :mod:`packaging`, to allow developers to make far-reaching changes without |
| 1210 | being constrained by backward compatibility. |
Éric Araujo | 765e94f | 2011-06-03 17:26:59 +0200 | [diff] [blame] | 1211 | :mod:`distutils` is still provided in the standard library, but users are |
| 1212 | encouraged to transition to :mod:`packaging`. For older versions of Python, a |
Éric Araujo | 4f61a2d | 2012-04-04 23:01:01 -0400 | [diff] [blame] | 1213 | backport compatible with Python 2.5 and newer and 3.2 is available on PyPI |
| 1214 | under the name `distutils2 <http://pypi.python.org/pypi/Distutils2>`_. |
Éric Araujo | 765e94f | 2011-06-03 17:26:59 +0200 | [diff] [blame] | 1215 | |
| 1216 | .. TODO add examples and howto to the packaging docs and link to them |
| 1217 | |
| 1218 | |
Georg Brandl | 4c7c3c5 | 2012-03-10 22:36:48 +0100 | [diff] [blame] | 1219 | pdb |
| 1220 | --- |
| 1221 | |
| 1222 | * Tab-completion is now available not only for command names, but also their |
| 1223 | arguments. For example, for the ``break`` command, function and file names |
| 1224 | are completed. (Contributed by Georg Brandl in :issue:`14210`) |
| 1225 | |
| 1226 | |
Antoine Pitrou | 9a86447 | 2012-05-04 23:15:47 +0200 | [diff] [blame] | 1227 | pickle |
| 1228 | ------ |
| 1229 | |
| 1230 | :class:`pickle.Pickler` objects now have an optional |
| 1231 | :attr:`~pickle.Pickler.dispatch_table` attribute allowing to set per-pickler |
| 1232 | reduction functions. |
| 1233 | (Contributed by Richard Oudkerk in :issue:`14166`.) |
| 1234 | |
| 1235 | |
Victor Stinner | 383c3fc | 2011-05-25 01:35:05 +0200 | [diff] [blame] | 1236 | pydoc |
| 1237 | ----- |
| 1238 | |
Victor Stinner | 6daa33c | 2011-05-25 01:41:22 +0200 | [diff] [blame] | 1239 | The Tk GUI and the :func:`~pydoc.serve` function have been removed from the |
| 1240 | :mod:`pydoc` module: ``pydoc -g`` and :func:`~pydoc.serve` have been deprecated |
| 1241 | in Python 3.2. |
Victor Stinner | 383c3fc | 2011-05-25 01:35:05 +0200 | [diff] [blame] | 1242 | |
| 1243 | |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1244 | sched |
| 1245 | ----- |
Victor Stinner | 754851f | 2011-04-19 23:58:51 +0200 | [diff] [blame] | 1246 | |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1247 | * :meth:`~sched.scheduler.run` now accepts a *blocking* parameter which when |
| 1248 | set to False makes the method execute the scheduled events due to expire |
| 1249 | soonest (if any) and then return immediately. |
| 1250 | This is useful in case you want to use the :class:`~sched.scheduler` in |
| 1251 | non-blocking applications. (Contributed by Giampaolo Rodolà in :issue:`13449`) |
Victor Stinner | 754851f | 2011-04-19 23:58:51 +0200 | [diff] [blame] | 1252 | |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1253 | * :class:`~sched.scheduler` class can now be safely used in multi-threaded |
| 1254 | environments. (Contributed by Josiah Carlson and Giampaolo Rodolà in |
| 1255 | :issue:`8684`) |
| 1256 | |
| 1257 | * *timefunc* and *delayfunct* parameters of :class:`~sched.scheduler` class |
| 1258 | constructor are now optional and defaults to :func:`time.time` and |
| 1259 | :func:`time.sleep` respectively. (Contributed by Chris Clark in |
| 1260 | :issue:`13245`) |
| 1261 | |
| 1262 | * :meth:`~sched.scheduler.enter` and :meth:`~sched.scheduler.enterabs` |
| 1263 | *argument* parameter is now optional. (Contributed by Chris Clark in |
| 1264 | :issue:`13245`) |
| 1265 | |
| 1266 | * :meth:`~sched.scheduler.enter` and :meth:`~sched.scheduler.enterabs` |
| 1267 | now accept a *kwargs* parameter. (Contributed by Chris Clark in |
| 1268 | :issue:`13245`) |
| 1269 | |
| 1270 | |
| 1271 | shutil |
| 1272 | ------ |
| 1273 | |
| 1274 | * The :mod:`shutil` module has these new fuctions: |
| 1275 | |
| 1276 | * :func:`~shutil.disk_usage`: provides total, used and free disk space |
| 1277 | statistics. (Contributed by Giampaolo Rodolà in :issue:`12442`) |
| 1278 | * :func:`~shutil.chown`: allows one to change user and/or group of the given |
| 1279 | path also specifying the user/group names and not only their numeric |
| 1280 | ids. (Contributed by Sandro Tosi in :issue:`12191`) |
Victor Stinner | a929335 | 2011-04-30 15:21:58 +0200 | [diff] [blame] | 1281 | |
Antoine Pitrou | 9a86447 | 2012-05-04 23:15:47 +0200 | [diff] [blame] | 1282 | * The new :func:`shutil.get_terminal_size` function returns the size of the |
| 1283 | terminal window the interpreter is attached to. |
| 1284 | (Contributed by Zbigniew Jędrzejewski-Szmek in :issue:`13609`.) |
| 1285 | |
| 1286 | * Several functions now take an optional ``symlinks`` argument: when that |
| 1287 | parameter is true, symlinks aren't dereferenced and the operation instead |
| 1288 | acts on the symlink itself (or creates one, if relevant). |
| 1289 | (Contributed by Hynek Schlawack in :issue:`12715`.) |
| 1290 | |
| 1291 | |
Victor Stinner | fa0e3d5 | 2011-05-09 01:01:09 +0200 | [diff] [blame] | 1292 | |
Victor Stinner | a929335 | 2011-04-30 15:21:58 +0200 | [diff] [blame] | 1293 | signal |
| 1294 | ------ |
| 1295 | |
Victor Stinner | fa0e3d5 | 2011-05-09 01:01:09 +0200 | [diff] [blame] | 1296 | * The :mod:`signal` module has new functions: |
Victor Stinner | a929335 | 2011-04-30 15:21:58 +0200 | [diff] [blame] | 1297 | |
Victor Stinner | b3e7219 | 2011-05-08 01:46:11 +0200 | [diff] [blame] | 1298 | * :func:`~signal.pthread_sigmask`: fetch and/or change the signal mask of the |
| 1299 | calling thread (Contributed by Jean-Paul Calderone in :issue:`8407`) ; |
| 1300 | * :func:`~signal.pthread_kill`: send a signal to a thread ; |
| 1301 | * :func:`~signal.sigpending`: examine pending functions ; |
| 1302 | * :func:`~signal.sigwait`: wait a signal. |
Ross Lagerwall | bc80822 | 2011-06-25 12:13:40 +0200 | [diff] [blame] | 1303 | * :func:`~signal.sigwaitinfo`: wait for a signal, returning detailed |
| 1304 | information about it. |
| 1305 | * :func:`~signal.sigtimedwait`: like :func:`~signal.sigwaitinfo` but with a |
| 1306 | timeout. |
Victor Stinner | a929335 | 2011-04-30 15:21:58 +0200 | [diff] [blame] | 1307 | |
Victor Stinner | d49b1f1 | 2011-05-08 02:03:15 +0200 | [diff] [blame] | 1308 | * The signal handler writes the signal number as a single byte instead of |
| 1309 | a nul byte into the wakeup file descriptor. So it is possible to wait more |
| 1310 | than one signal and know which signals were raised. |
| 1311 | |
Victor Stinner | 388196e | 2011-05-10 17:13:00 +0200 | [diff] [blame] | 1312 | * :func:`signal.signal` and :func:`signal.siginterrupt` raise an OSError, |
| 1313 | instead of a RuntimeError: OSError has an errno attribute. |
| 1314 | |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1315 | smtplib |
| 1316 | ------- |
| 1317 | |
| 1318 | The :class:`~smtplib.SMTP_SSL` constructor and the :meth:`~smtplib.SMTP.starttls` |
| 1319 | method now accept an SSLContext parameter to control parameters of the secure |
| 1320 | channel. |
| 1321 | |
| 1322 | (Contributed by Kasun Herath in :issue:`8809`) |
| 1323 | |
| 1324 | |
Nick Coghlan | 96fe56a | 2011-08-22 11:55:57 +1000 | [diff] [blame] | 1325 | socket |
| 1326 | ------ |
| 1327 | |
Charles-François Natali | 47413c1 | 2011-10-06 19:47:44 +0200 | [diff] [blame] | 1328 | * The :class:`~socket.socket` class now exposes additional methods to process |
| 1329 | ancillary data when supported by the underlying platform: |
Nick Coghlan | 96fe56a | 2011-08-22 11:55:57 +1000 | [diff] [blame] | 1330 | |
Charles-François Natali | 47413c1 | 2011-10-06 19:47:44 +0200 | [diff] [blame] | 1331 | * :func:`~socket.socket.sendmsg` |
| 1332 | * :func:`~socket.socket.recvmsg` |
| 1333 | * :func:`~socket.socket.recvmsg_into` |
Nick Coghlan | 96fe56a | 2011-08-22 11:55:57 +1000 | [diff] [blame] | 1334 | |
Charles-François Natali | 47413c1 | 2011-10-06 19:47:44 +0200 | [diff] [blame] | 1335 | (Contributed by David Watson in :issue:`6560`, based on an earlier patch by |
| 1336 | Heiko Wundram) |
| 1337 | |
| 1338 | * The :class:`~socket.socket` class now supports the PF_CAN protocol family |
| 1339 | (http://en.wikipedia.org/wiki/Socketcan), on Linux |
| 1340 | (http://lwn.net/Articles/253425). |
| 1341 | |
| 1342 | (Contributed by Matthias Fuchs, updated by Tiago Gonçalves in :issue:`10141`) |
| 1343 | |
Charles-François Natali | 10b8cf4 | 2011-11-10 19:21:37 +0100 | [diff] [blame] | 1344 | * The :class:`~socket.socket` class now supports the PF_RDS protocol family |
| 1345 | (http://en.wikipedia.org/wiki/Reliable_Datagram_Sockets and |
| 1346 | http://oss.oracle.com/projects/rds/). |
Victor Stinner | 754851f | 2011-04-19 23:58:51 +0200 | [diff] [blame] | 1347 | |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1348 | |
Victor Stinner | 99c8b16 | 2011-05-24 12:05:19 +0200 | [diff] [blame] | 1349 | ssl |
| 1350 | --- |
| 1351 | |
Antoine Pitrou | 2c0a967 | 2011-11-17 02:09:13 +0100 | [diff] [blame] | 1352 | * The :mod:`ssl` module has two new random generation functions: |
Victor Stinner | 99c8b16 | 2011-05-24 12:05:19 +0200 | [diff] [blame] | 1353 | |
| 1354 | * :func:`~ssl.RAND_bytes`: generate cryptographically strong |
| 1355 | pseudo-random bytes. |
| 1356 | * :func:`~ssl.RAND_pseudo_bytes`: generate pseudo-random bytes. |
| 1357 | |
Antoine Pitrou | 2c0a967 | 2011-11-17 02:09:13 +0100 | [diff] [blame] | 1358 | (Contributed by Victor Stinner in :issue:`12049`) |
| 1359 | |
| 1360 | * The :mod:`ssl` module now exposes a finer-grained exception hierarchy |
| 1361 | in order to make it easier to inspect the various kinds of errors. |
| 1362 | |
| 1363 | (Contributed by Antoine Pitrou in :issue:`11183`) |
| 1364 | |
| 1365 | * :meth:`~ssl.SSLContext.load_cert_chain` now accepts a *password* argument |
| 1366 | to be used if the private key is encrypted. |
| 1367 | |
| 1368 | (Contributed by Adam Simpkins in :issue:`12803`) |
| 1369 | |
Antoine Pitrou | 73fc814 | 2011-12-23 20:58:36 +0100 | [diff] [blame] | 1370 | * Diffie-Hellman key exchange, both regular and Elliptic Curve-based, is |
| 1371 | now supported through the :meth:`~ssl.SSLContext.load_dh_params` and |
| 1372 | :meth:`~ssl.SSLContext.set_ecdh_curve` methods. |
| 1373 | |
| 1374 | (Contributed by Antoine Pitrou in :issue:`13626` and :issue:`13627`) |
| 1375 | |
Antoine Pitrou | 2c0a967 | 2011-11-17 02:09:13 +0100 | [diff] [blame] | 1376 | * SSL sockets have a new :meth:`~ssl.SSLSocket.get_channel_binding` method |
| 1377 | allowing the implementation of certain authentication mechanisms such as |
| 1378 | SCRAM-SHA-1-PLUS. |
| 1379 | |
| 1380 | (Contributed by Jacek Konieczny in :issue:`12551`) |
| 1381 | |
Antoine Pitrou | 73fc814 | 2011-12-23 20:58:36 +0100 | [diff] [blame] | 1382 | * You can query the SSL compression algorithm used by an SSL socket, thanks |
| 1383 | to its new :meth:`~ssl.SSLSocket.compression` method. |
| 1384 | |
| 1385 | (Contributed by Antoine Pitrou in :issue:`13634`) |
| 1386 | |
Antoine Pitrou | 9a86447 | 2012-05-04 23:15:47 +0200 | [diff] [blame] | 1387 | * Support has been added for the Next Procotol Negotiation extension using |
| 1388 | the :meth:`ssl.SSLContext.set_npn_protocols` method. |
| 1389 | |
| 1390 | (Contributed by Colin Marc in :issue:`14204`) |
| 1391 | |
Giampaolo Rodola' | ffa1d0b | 2012-05-15 15:30:25 +0200 | [diff] [blame] | 1392 | stat |
| 1393 | ---- |
| 1394 | |
| 1395 | - The undocumented tarfile.filemode function has been moved to |
| 1396 | :func:`stat.filemode`. It can be used to convert a file's mode to a string of |
| 1397 | the form '-rwxrwxrwx'. |
| 1398 | |
| 1399 | (Contributed by Giampaolo Rodolà in :issue:`14807`) |
Antoine Pitrou | 73fc814 | 2011-12-23 20:58:36 +0100 | [diff] [blame] | 1400 | |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1401 | sys |
| 1402 | --- |
Giampaolo Rodola' | 210e7ca | 2011-07-01 13:55:36 +0200 | [diff] [blame] | 1403 | |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1404 | * The :mod:`sys` module has a new :data:`~sys.thread_info` :term:`struct |
| 1405 | sequence` holding informations about the thread implementation. |
Giampaolo Rodola' | 210e7ca | 2011-07-01 13:55:36 +0200 | [diff] [blame] | 1406 | |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1407 | (:issue:`11223`) |
Giampaolo Rodola' | 096dcb1 | 2011-06-27 11:17:51 +0200 | [diff] [blame] | 1408 | |
Nick Coghlan | 4fae8cd | 2012-06-11 23:07:51 +1000 | [diff] [blame^] | 1409 | textwrap |
| 1410 | -------- |
| 1411 | |
| 1412 | * The :mod:`textwrap` module has a new :func:`~textwrap.indent` that makes |
| 1413 | it straightforward to add a common prefix to selected lines in a block |
| 1414 | of text. |
| 1415 | |
| 1416 | (:issue:`13857`) |
Antoine Pitrou | 5a8bc6f | 2011-11-17 02:20:48 +0100 | [diff] [blame] | 1417 | |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1418 | time |
| 1419 | ---- |
Antoine Pitrou | 5a8bc6f | 2011-11-17 02:20:48 +0100 | [diff] [blame] | 1420 | |
Victor Stinner | ec89539 | 2012-04-29 02:41:27 +0200 | [diff] [blame] | 1421 | The :pep:`418` added new functions to the :mod:`time` module: |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1422 | |
Victor Stinner | ec89539 | 2012-04-29 02:41:27 +0200 | [diff] [blame] | 1423 | * :func:`~time.get_clock_info`: Get information on a clock. |
| 1424 | * :func:`~time.monotonic`: Monotonic clock (cannot go backward), not affected |
| 1425 | by system clock updates. |
| 1426 | * :func:`~time.perf_counter`: Performance counter with the highest available |
| 1427 | resolution to measure a short duration. |
| 1428 | * :func:`~time.process_time`: Sum of the system and user CPU time of the |
| 1429 | current process. |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1430 | |
Victor Stinner | ec89539 | 2012-04-29 02:41:27 +0200 | [diff] [blame] | 1431 | Other new functions: |
| 1432 | |
| 1433 | * :func:`~time.clock_getres`, :func:`~time.clock_gettime` and |
| 1434 | :func:`~time.clock_settime` functions with ``CLOCK_xxx`` constants. |
| 1435 | (Contributed by Victor Stinner in :issue:`10278`) |
Victor Stinner | f4c54ff | 2012-02-08 01:48:34 +0100 | [diff] [blame] | 1436 | |
Antoine Pitrou | 5a8bc6f | 2011-11-17 02:20:48 +0100 | [diff] [blame] | 1437 | |
Victor Stinner | 0db176f | 2012-04-16 00:16:30 +0200 | [diff] [blame] | 1438 | types |
| 1439 | ----- |
| 1440 | |
| 1441 | Add a new :class:`types.MappingProxyType` class: Read-only proxy of a mapping. |
| 1442 | (:issue:`14386`) |
| 1443 | |
| 1444 | |
Nick Coghlan | 7fc570a | 2012-05-20 02:34:13 +1000 | [diff] [blame] | 1445 | The new functions `types.new_class` and `types.prepare_class` provide support |
| 1446 | for PEP 3115 compliant dynamic type creation. (:issue:`14588`) |
| 1447 | |
| 1448 | |
Senthil Kumaran | de49d64 | 2011-10-16 23:54:44 +0800 | [diff] [blame] | 1449 | urllib |
| 1450 | ------ |
| 1451 | |
| 1452 | The :class:`~urllib.request.Request` class, now accepts a *method* argument |
| 1453 | used by :meth:`~urllib.request.Request.get_method` to determine what HTTP method |
Senthil Kumaran | a41c942 | 2011-10-20 02:37:08 +0800 | [diff] [blame] | 1454 | should be used. For example, this will send a ``'HEAD'`` request:: |
Senthil Kumaran | de49d64 | 2011-10-16 23:54:44 +0800 | [diff] [blame] | 1455 | |
| 1456 | >>> urlopen(Request('http://www.python.org', method='HEAD')) |
| 1457 | |
| 1458 | (:issue:`1673007`) |
Giampaolo Rodola' | 096dcb1 | 2011-06-27 11:17:51 +0200 | [diff] [blame] | 1459 | |
Giampaolo Rodola' | be55d99 | 2011-11-22 13:33:34 +0100 | [diff] [blame] | 1460 | |
Éric Araujo | 4f61a2d | 2012-04-04 23:01:01 -0400 | [diff] [blame] | 1461 | webbrowser |
| 1462 | ---------- |
| 1463 | |
| 1464 | The :mod:`webbrowser` module supports more browsers: Google Chrome (named |
| 1465 | :program:`chrome`, :program:`chromium`, :program:`chrome-browser` or |
| 1466 | :program:`chromium-browser` depending on the version and operating system) as |
| 1467 | well as the the generic launchers :program:`xdg-open` from the FreeDesktop.org |
| 1468 | project and :program:`gvfs-open` which is the default URI handler for GNOME 3. |
| 1469 | |
| 1470 | (:issue:`13620` and :issue:`14493`) |
| 1471 | |
| 1472 | |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 1473 | Optimizations |
| 1474 | ============= |
| 1475 | |
| 1476 | Major performance enhancements have been added: |
| 1477 | |
Éric Araujo | 4f61a2d | 2012-04-04 23:01:01 -0400 | [diff] [blame] | 1478 | * Thanks to :pep:`393`, some operations on Unicode strings have been optimized: |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1479 | |
| 1480 | * the memory footprint is divided by 2 to 4 depending on the text |
Victor Stinner | a996f1e | 2011-11-21 13:14:43 +0100 | [diff] [blame] | 1481 | * encode an ASCII string to UTF-8 doesn't need to encode characters anymore, |
| 1482 | the UTF-8 representation is shared with the ASCII representation |
Victor Stinner | 6099a03 | 2011-12-18 14:22:26 +0100 | [diff] [blame] | 1483 | * the UTF-8 encoder has been optimized |
| 1484 | * repeating a single ASCII letter and getting a substring of a ASCII strings |
| 1485 | is 4 times faster |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 1486 | |
Antoine Pitrou | 5cec9d2 | 2012-05-17 17:37:02 +0200 | [diff] [blame] | 1487 | * UTF-8 and UTF-16 decoding is now 2x to 4x faster. |
| 1488 | |
| 1489 | (contributed by Serhiy Storchaka, :issue:`14624` and :issue:`14738`.) |
| 1490 | |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 1491 | |
| 1492 | Build and C API Changes |
| 1493 | ======================= |
| 1494 | |
| 1495 | Changes to Python's build process and to the C API include: |
| 1496 | |
Stefan Krah | 95b1ba6 | 2012-02-29 17:27:21 +0100 | [diff] [blame] | 1497 | * New :pep:`3118` related function: |
| 1498 | |
| 1499 | * :c:func:`PyMemoryView_FromMemory` |
| 1500 | |
Éric Araujo | 4f61a2d | 2012-04-04 23:01:01 -0400 | [diff] [blame] | 1501 | * :pep:`393` added new Unicode types, macros and functions: |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1502 | |
Victor Stinner | a996f1e | 2011-11-21 13:14:43 +0100 | [diff] [blame] | 1503 | * High-level API: |
| 1504 | |
| 1505 | * :c:func:`PyUnicode_CopyCharacters` |
| 1506 | * :c:func:`PyUnicode_FindChar` |
| 1507 | * :c:func:`PyUnicode_GetLength`, :c:macro:`PyUnicode_GET_LENGTH` |
| 1508 | * :c:func:`PyUnicode_New` |
| 1509 | * :c:func:`PyUnicode_Substring` |
| 1510 | * :c:func:`PyUnicode_ReadChar`, :c:func:`PyUnicode_WriteChar` |
| 1511 | |
| 1512 | * Low-level API: |
| 1513 | |
| 1514 | * :c:type:`Py_UCS1`, :c:type:`Py_UCS2`, :c:type:`Py_UCS4` types |
| 1515 | * :c:type:`PyASCIIObject` and :c:type:`PyCompactUnicodeObject` structures |
| 1516 | * :c:macro:`PyUnicode_READY` |
| 1517 | * :c:func:`PyUnicode_FromKindAndData` |
| 1518 | * :c:func:`PyUnicode_AsUCS4`, :c:func:`PyUnicode_AsUCS4Copy` |
| 1519 | * :c:macro:`PyUnicode_DATA`, :c:macro:`PyUnicode_1BYTE_DATA`, |
| 1520 | :c:macro:`PyUnicode_2BYTE_DATA`, :c:macro:`PyUnicode_4BYTE_DATA` |
| 1521 | * :c:macro:`PyUnicode_KIND` with :c:type:`PyUnicode_Kind` enum: |
| 1522 | :c:data:`PyUnicode_WCHAR_KIND`, :c:data:`PyUnicode_1BYTE_KIND`, |
| 1523 | :c:data:`PyUnicode_2BYTE_KIND`, :c:data:`PyUnicode_4BYTE_KIND` |
| 1524 | * :c:macro:`PyUnicode_READ`, :c:macro:`PyUnicode_READ_CHAR`, :c:macro:`PyUnicode_WRITE` |
| 1525 | * :c:macro:`PyUnicode_MAX_CHAR_VALUE` |
| 1526 | |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 1527 | |
| 1528 | |
Victor Stinner | d1be878 | 2011-12-09 00:10:41 +0100 | [diff] [blame] | 1529 | Deprecated |
| 1530 | ========== |
| 1531 | |
Georg Brandl | 0cd25c9 | 2011-04-29 13:45:54 +0200 | [diff] [blame] | 1532 | Unsupported Operating Systems |
Victor Stinner | d1be878 | 2011-12-09 00:10:41 +0100 | [diff] [blame] | 1533 | ----------------------------- |
Victor Stinner | b90db4c | 2011-04-26 22:48:24 +0200 | [diff] [blame] | 1534 | |
Brian Curtin | 49a40cd | 2011-05-02 22:30:06 -0500 | [diff] [blame] | 1535 | OS/2 and VMS are no longer supported due to the lack of a maintainer. |
| 1536 | |
| 1537 | Windows 2000 and Windows platforms which set ``COMSPEC`` to ``command.com`` |
| 1538 | are no longer supported due to maintenance burden. |
Victor Stinner | b90db4c | 2011-04-26 22:48:24 +0200 | [diff] [blame] | 1539 | |
| 1540 | |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1541 | Deprecated Python modules, functions and methods |
Victor Stinner | d1be878 | 2011-12-09 00:10:41 +0100 | [diff] [blame] | 1542 | ------------------------------------------------ |
Victor Stinner | 19bd069 | 2011-11-16 00:18:57 +0100 | [diff] [blame] | 1543 | |
Éric Araujo | 4f61a2d | 2012-04-04 23:01:01 -0400 | [diff] [blame] | 1544 | * The :mod:`distutils` module has been deprecated. Use the new |
R David Murray | 4a1ad91 | 2012-03-26 13:34:46 -0400 | [diff] [blame] | 1545 | :mod:`packaging` module instead. |
Victor Stinner | 19bd069 | 2011-11-16 00:18:57 +0100 | [diff] [blame] | 1546 | * The ``unicode_internal`` codec has been deprecated because of the |
Sandro Tosi | cd89912 | 2012-01-22 12:16:04 +0100 | [diff] [blame] | 1547 | :pep:`393`, use UTF-8, UTF-16 (``utf-16-le`` or ``utf-16-be``), or UTF-32 |
| 1548 | (``utf-32-le`` or ``utf-32-be``) |
Victor Stinner | 19bd069 | 2011-11-16 00:18:57 +0100 | [diff] [blame] | 1549 | * :meth:`ftplib.FTP.nlst` and :meth:`ftplib.FTP.dir`: use |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1550 | :meth:`ftplib.FTP.mlsd` |
Victor Stinner | 19bd069 | 2011-11-16 00:18:57 +0100 | [diff] [blame] | 1551 | * :func:`platform.popen`: use the :mod:`subprocess` module. Check especially |
| 1552 | the :ref:`subprocess-replacements` section. |
| 1553 | * :issue:`13374`: The Windows bytes API has been deprecated in the :mod:`os` |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1554 | module. Use Unicode filenames, instead of bytes filenames, to not depend on |
Victor Stinner | 19bd069 | 2011-11-16 00:18:57 +0100 | [diff] [blame] | 1555 | the ANSI code page anymore and to support any filename. |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 1556 | * :issue:`13988`: The :mod:`xml.etree.cElementTree` module is deprecated. The |
| 1557 | accelerator is used automatically whenever available. |
Victor Stinner | 47620a6 | 2012-04-29 02:52:39 +0200 | [diff] [blame] | 1558 | * The behaviour of :func:`time.clock` depends on the platform: use the new |
| 1559 | :func:`time.perf_counter` or :func:`time.process_time` function instead, |
| 1560 | depending on your requirements, to have a well defined behaviour. |
Victor Stinner | 19bd069 | 2011-11-16 00:18:57 +0100 | [diff] [blame] | 1561 | |
| 1562 | |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1563 | Deprecated functions and types of the C API |
Victor Stinner | d1be878 | 2011-12-09 00:10:41 +0100 | [diff] [blame] | 1564 | ------------------------------------------- |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1565 | |
Éric Araujo | 4f61a2d | 2012-04-04 23:01:01 -0400 | [diff] [blame] | 1566 | The :c:type:`Py_UNICODE` has been deprecated by :pep:`393` and will be |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1567 | removed in Python 4. All functions using this type are deprecated: |
| 1568 | |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1569 | Unicode functions and methods using :c:type:`Py_UNICODE` and |
| 1570 | :c:type:`Py_UNICODE*` types: |
| 1571 | |
| 1572 | * :c:macro:`PyUnicode_FromUnicode`: use :c:func:`PyUnicode_FromWideChar` or |
| 1573 | :c:func:`PyUnicode_FromKindAndData` |
| 1574 | * :c:macro:`PyUnicode_AS_UNICODE`, :c:func:`PyUnicode_AsUnicode`, |
| 1575 | :c:func:`PyUnicode_AsUnicodeAndSize`: use :c:func:`PyUnicode_AsWideCharString` |
| 1576 | * :c:macro:`PyUnicode_AS_DATA`: use :c:macro:`PyUnicode_DATA` with |
| 1577 | :c:macro:`PyUnicode_READ` and :c:macro:`PyUnicode_WRITE` |
| 1578 | * :c:macro:`PyUnicode_GET_SIZE`, :c:func:`PyUnicode_GetSize`: use |
| 1579 | :c:macro:`PyUnicode_GET_LENGTH` or :c:func:`PyUnicode_GetLength` |
| 1580 | * :c:macro:`PyUnicode_GET_DATA_SIZE`: use |
| 1581 | ``PyUnicode_GET_LENGTH(str) * PyUnicode_KIND(str)`` (only work on ready |
| 1582 | strings) |
Victor Stinner | bf6e560 | 2011-12-12 01:53:47 +0100 | [diff] [blame] | 1583 | * :c:func:`PyUnicode_AsUnicodeCopy`: use :c:func:`PyUnicode_AsUCS4Copy` or |
| 1584 | :c:func:`PyUnicode_AsWideCharString` |
Victor Stinner | ab59594 | 2011-12-17 04:59:06 +0100 | [diff] [blame] | 1585 | * :c:func:`PyUnicode_GetMax` |
| 1586 | |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1587 | |
Victor Stinner | a996f1e | 2011-11-21 13:14:43 +0100 | [diff] [blame] | 1588 | Functions and macros manipulating Py_UNICODE* strings: |
| 1589 | |
| 1590 | * :c:macro:`Py_UNICODE_strlen`: use :c:func:`PyUnicode_GetLength` or |
| 1591 | :c:macro:`PyUnicode_GET_LENGTH` |
| 1592 | * :c:macro:`Py_UNICODE_strcat`: use :c:func:`PyUnicode_CopyCharacters` or |
| 1593 | :c:func:`PyUnicode_FromFormat` |
| 1594 | * :c:macro:`Py_UNICODE_strcpy`, :c:macro:`Py_UNICODE_strncpy`, |
| 1595 | :c:macro:`Py_UNICODE_COPY`: use :c:func:`PyUnicode_CopyCharacters` or |
| 1596 | :c:func:`PyUnicode_Substring` |
| 1597 | * :c:macro:`Py_UNICODE_strcmp`: use :c:func:`PyUnicode_Compare` |
| 1598 | * :c:macro:`Py_UNICODE_strncmp`: use :c:func:`PyUnicode_Tailmatch` |
| 1599 | * :c:macro:`Py_UNICODE_strchr`, :c:macro:`Py_UNICODE_strrchr`: use |
| 1600 | :c:func:`PyUnicode_FindChar` |
Victor Stinner | 606e19d | 2012-01-04 03:59:16 +0100 | [diff] [blame] | 1601 | * :c:macro:`Py_UNICODE_FILL`: use :c:func:`PyUnicode_Fill` |
Victor Stinner | ab59594 | 2011-12-17 04:59:06 +0100 | [diff] [blame] | 1602 | * :c:macro:`Py_UNICODE_MATCH` |
Victor Stinner | a996f1e | 2011-11-21 13:14:43 +0100 | [diff] [blame] | 1603 | |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1604 | Encoders: |
| 1605 | |
| 1606 | * :c:func:`PyUnicode_Encode`: use :c:func:`PyUnicode_AsEncodedObject` |
| 1607 | * :c:func:`PyUnicode_EncodeUTF7` |
Victor Stinner | a996f1e | 2011-11-21 13:14:43 +0100 | [diff] [blame] | 1608 | * :c:func:`PyUnicode_EncodeUTF8`: use :c:func:`PyUnicode_AsUTF8` or |
| 1609 | :c:func:`PyUnicode_AsUTF8String` |
Victor Stinner | 46606ce | 2011-11-20 18:27:55 +0100 | [diff] [blame] | 1610 | * :c:func:`PyUnicode_EncodeUTF32` |
| 1611 | * :c:func:`PyUnicode_EncodeUTF16` |
| 1612 | * :c:func:`PyUnicode_EncodeUnicodeEscape:` use |
| 1613 | :c:func:`PyUnicode_AsUnicodeEscapeString` |
| 1614 | * :c:func:`PyUnicode_EncodeRawUnicodeEscape:` use |
| 1615 | :c:func:`PyUnicode_AsRawUnicodeEscapeString` |
| 1616 | * :c:func:`PyUnicode_EncodeLatin1`: use :c:func:`PyUnicode_AsLatin1String` |
| 1617 | * :c:func:`PyUnicode_EncodeASCII`: use :c:func:`PyUnicode_AsASCIIString` |
| 1618 | * :c:func:`PyUnicode_EncodeCharmap` |
| 1619 | * :c:func:`PyUnicode_TranslateCharmap` |
| 1620 | * :c:func:`PyUnicode_EncodeMBCS`: use :c:func:`PyUnicode_AsMBCSString` or |
| 1621 | :c:func:`PyUnicode_EncodeCodePage` (with ``CP_ACP`` code_page) |
| 1622 | * :c:func:`PyUnicode_EncodeDecimal`, |
| 1623 | :c:func:`PyUnicode_TransformDecimalToASCII` |
| 1624 | |
| 1625 | |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 1626 | Porting to Python 3.3 |
| 1627 | ===================== |
| 1628 | |
| 1629 | This section lists previously described changes and other bugfixes |
Antoine Pitrou | 037ffbf | 2011-10-24 00:25:41 +0200 | [diff] [blame] | 1630 | that may require changes to your code. |
| 1631 | |
| 1632 | Porting Python code |
| 1633 | ------------------- |
Giampaolo Rodolà | 3108f98 | 2011-02-24 20:59:48 +0000 | [diff] [blame] | 1634 | |
Georg Brandl | d6c4340 | 2012-03-07 08:55:52 +0100 | [diff] [blame] | 1635 | .. XXX add a point about hash randomization and that it's always on in 3.3 |
| 1636 | |
Victor Stinner | 19bd069 | 2011-11-16 00:18:57 +0100 | [diff] [blame] | 1637 | * :issue:`12326`: On Linux, sys.platform doesn't contain the major version |
Victor Stinner | ff3d939 | 2011-08-20 23:39:26 +0200 | [diff] [blame] | 1638 | anymore. It is now always 'linux', instead of 'linux2' or 'linux3' depending |
| 1639 | on the Linux version used to build Python. Replace sys.platform == 'linux2' |
| 1640 | with sys.platform.startswith('linux'), or directly sys.platform == 'linux' if |
| 1641 | you don't need to support older Python versions. |
Éric Araujo | c09fca6 | 2011-03-23 02:06:24 +0100 | [diff] [blame] | 1642 | |
Victor Stinner | ecc6e66 | 2012-03-14 00:39:29 +0100 | [diff] [blame] | 1643 | * :issue:`13847`, :issue:`14180`: :mod:`time` and :mod:`datetime`: |
| 1644 | :exc:`OverflowError` is now raised instead of :exc:`ValueError` if a |
| 1645 | timestamp is out of range. :exc:`OSError` is now raised if C functions |
| 1646 | :c:func:`gmtime` or :c:func:`localtime` failed. |
| 1647 | |
Brett Cannon | c204348 | 2012-04-29 20:59:41 -0400 | [diff] [blame] | 1648 | * The default finders used by import now utilize a cache of what is contained |
| 1649 | within a specific directory. If you create a Python source file or sourceless |
| 1650 | bytecode file, make sure to call :func:`importlib.invalidate_caches` to clear |
| 1651 | out the cache for the finders to notice the new file. |
| 1652 | |
| 1653 | * :exc:`ImportError` now uses the full name of the module that was attemped to |
| 1654 | be imported. Doctests that check ImportErrors' message will need to be |
| 1655 | updated to use the full name of the module instead of just the tail of the |
| 1656 | name. |
| 1657 | |
| 1658 | * The **index** argument to :func:`__import__` now defaults to 0 instead of -1 |
| 1659 | and no longer support negative values. It was an oversight when :pep:`328` was |
| 1660 | implemented that the default value remained -1. If you need to continue to |
| 1661 | perform a relative import followed by an absolute import, then perform the |
| 1662 | relative import using an index of 1, followed by another import using an |
| 1663 | index of 0. It is preferred, though, that you use |
| 1664 | :func:`importlib.import_module` rather than call :func:`__import__` directly. |
| 1665 | |
| 1666 | * :func:`__import__` no longer allows one to use an index value other than 0 |
| 1667 | for top-level modules. E.g. ``__import__('sys', level=1)`` is now an error. |
| 1668 | |
| 1669 | * Because :attr:`sys.meta_path` and :attr:`sys.path_hooks` now have finders on |
| 1670 | them by default, you will most likely want to use :meth:`list.insert` instead |
| 1671 | of :meth:`list.append` to add to those lists. |
| 1672 | |
| 1673 | * Because ``None`` is now inserted into :attr:`sys.path_importer_cache`, if you |
| 1674 | are clearing out entries in the dictionary of paths that do not have a |
| 1675 | finder, you will need to remove keys paired with values of ``None`` **and** |
| 1676 | :class:`imp.NullImporter` to be backwards-compatible. This will need to extra |
| 1677 | overhead on older versions of Python that re-insert ``None`` into |
| 1678 | :attr:`sys.path_importer_cache` where it repesents the use of implicit |
| 1679 | finders, but semantically it should not change anything. |
| 1680 | |
| 1681 | * :meth:`importlib.abc.SourceLoader.path_mtime` is now deprecated in favour of |
| 1682 | :meth:`importlib.abc.SourceLoader.path_stats` as bytecode files now store |
| 1683 | both the modification time and size of the source file the bytecode file was |
| 1684 | compiled from. |
| 1685 | |
| 1686 | |
Antoine Pitrou | 037ffbf | 2011-10-24 00:25:41 +0200 | [diff] [blame] | 1687 | Porting C code |
| 1688 | -------------- |
| 1689 | |
Stefan Krah | 54c3203 | 2012-02-29 17:47:21 +0100 | [diff] [blame] | 1690 | * In the course of changes to the buffer API the undocumented |
| 1691 | :c:member:`~Py_buffer.smalltable` member of the |
| 1692 | :c:type:`Py_buffer` structure has been removed and the |
| 1693 | layout of the :c:type:`PyMemoryViewObject` has changed. |
| 1694 | |
| 1695 | All extensions relying on the relevant parts in ``memoryobject.h`` |
| 1696 | or ``object.h`` must be rebuilt. |
| 1697 | |
Antoine Pitrou | 037ffbf | 2011-10-24 00:25:41 +0200 | [diff] [blame] | 1698 | * Due to :ref:`PEP 393 <pep-393>`, the :c:type:`Py_UNICODE` type and all |
| 1699 | functions using this type are deprecated (but will stay available for |
| 1700 | at least five years). If you were using low-level Unicode APIs to |
| 1701 | construct and access unicode objects and you want to benefit of the |
Éric Araujo | 4f61a2d | 2012-04-04 23:01:01 -0400 | [diff] [blame] | 1702 | memory footprint reduction provided by PEP 393, you have to convert |
Antoine Pitrou | 037ffbf | 2011-10-24 00:25:41 +0200 | [diff] [blame] | 1703 | your code to the new :doc:`Unicode API <../c-api/unicode>`. |
| 1704 | |
| 1705 | However, if you only have been using high-level functions such as |
| 1706 | :c:func:`PyUnicode_Concat()`, :c:func:`PyUnicode_Join` or |
| 1707 | :c:func:`PyUnicode_FromFormat()`, your code will automatically take |
| 1708 | advantage of the new unicode representations. |
| 1709 | |
Antoine Pitrou | c229e6e | 2012-02-20 19:41:11 +0100 | [diff] [blame] | 1710 | Building C extensions |
| 1711 | --------------------- |
| 1712 | |
| 1713 | * The range of possible file names for C extensions has been narrowed. |
| 1714 | Very rarely used spellings have been suppressed: under POSIX, files |
| 1715 | named ``xxxmodule.so``, ``xxxmodule.abi3.so`` and |
| 1716 | ``xxxmodule.cpython-*.so`` are no longer recognized as implementing |
| 1717 | the ``xxx`` module. If you had been generating such files, you have |
| 1718 | to switch to the other spellings (i.e., remove the ``module`` string |
| 1719 | from the file names). |
| 1720 | |
| 1721 | (implemented in :issue:`14040`.) |
| 1722 | |
| 1723 | |
Antoine Pitrou | 037ffbf | 2011-10-24 00:25:41 +0200 | [diff] [blame] | 1724 | Other issues |
| 1725 | ------------ |
| 1726 | |
Éric Araujo | c09fca6 | 2011-03-23 02:06:24 +0100 | [diff] [blame] | 1727 | .. Issue #11591: When :program:`python` was started with :option:`-S`, |
| 1728 | ``import site`` will not add site-specific paths to the module search |
| 1729 | paths. In previous versions, it did. See changeset for doc changes in |
| 1730 | various files. Contributed by Carl Meyer with editions by Éric Araujo. |
Éric Araujo | be3bd57 | 2011-03-26 01:55:15 +0100 | [diff] [blame] | 1731 | |
Éric Araujo | bfc9729 | 2011-11-14 18:18:15 +0100 | [diff] [blame] | 1732 | .. Issue #10998: the -Q command-line flag and related artifacts have been |
Éric Araujo | be3bd57 | 2011-03-26 01:55:15 +0100 | [diff] [blame] | 1733 | removed. Code checking sys.flags.division_warning will need updating. |
| 1734 | Contributed by Éric Araujo. |