Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1 | +++++++++++ |
| 2 | Python News |
| 3 | +++++++++++ |
| 4 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 5 | What's New in Python 2.3 alpha 1? |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 6 | ================================= |
| 7 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 8 | *XXX Release date: DD-MMM-2002 XXX* |
| 9 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 10 | Type/class unification and new-style classes |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 11 | -------------------------------------------- |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 12 | |
Guido van Rossum | e343878 | 2002-08-12 03:42:03 +0000 | [diff] [blame] | 13 | - Assignment to __class__ is disallowed if either the old and the new |
| 14 | class is a statically allocated type object (such as defined by an |
Raymond Hettinger | f2f2a2c | 2002-08-19 14:25:03 +0000 | [diff] [blame] | 15 | extension module). This prevents anomalies like 2.__class__ = bool. |
Guido van Rossum | e343878 | 2002-08-12 03:42:03 +0000 | [diff] [blame] | 16 | |
| 17 | - New-style object creation and deallocation have been sped up |
| 18 | significantly; they are now faster than classic instance creation |
| 19 | and deallocation. |
| 20 | |
| 21 | - The __slots__ variable can now mention "private" names, and the |
| 22 | right thing will happen (e.g. __slots__ = ["__foo"]). |
| 23 | |
| 24 | - The built-ins slice() and buffer() are now callable types. The |
| 25 | types classobj (formerly class), code, function, instance, and |
| 26 | instancemethod (formerly instance-method), which have no built-in |
| 27 | names but are accessible through the types module, are now also |
| 28 | callable. The type dict-proxy is renamed to dictproxy. |
| 29 | |
| 30 | - Cycles going through the __class__ link of a new-style instance are |
| 31 | now detected by the garbage collector. |
| 32 | |
| 33 | - Classes using __slots__ are now properly garbage collected. |
| 34 | [SF bug 519621] |
| 35 | |
| 36 | - Tightened the __slots__ rules: a slot name must be a valid Python |
| 37 | identifier. |
| 38 | |
| 39 | - The constructor for the module type now requires a name argument and |
| 40 | takes an optional docstring argument. Previously, this constructor |
| 41 | ignored its arguments. As a consequence, deriving a class from a |
| 42 | module (not from the module type) is now illegal; previously this |
| 43 | created an unnamed module, just like invoking the module type did. |
| 44 | [SF bug 563060] |
| 45 | |
| 46 | - A new type object, 'basestring', is added. This is a common base type |
| 47 | for 'str' and 'unicode', and can be used instead of |
| 48 | types.StringTypes, e.g. to test whether something is "a string": |
| 49 | isinstance(x, basestring) is True for Unicode and 8-bit strings. This |
| 50 | is an abstract base class and cannot be instantiated directly. |
| 51 | |
| 52 | - Changed new-style class instantiation so that when C's __new__ |
| 53 | method returns something that's not a C instance, its __init__ is |
| 54 | not called. [SF bug #537450] |
| 55 | |
| 56 | - Fixed super() to work correctly with class methods. [SF bug #535444] |
| 57 | |
| 58 | - If you try to pickle an instance of a class that has __slots__ but |
| 59 | doesn't define or override __getstate__, a TypeError is now raised. |
| 60 | This is done by adding a bozo __getstate__ to the class that always |
| 61 | raises TypeError. (Before, this would appear to be pickled, but the |
| 62 | state of the slots would be lost.) |
| 63 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 64 | Core and builtins |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 65 | ----------------- |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 66 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 67 | - Codec error handling callbacks (PEP 293) are implemented. |
| 68 | Error handling in unicode.encode or str.decode can now be customized. |
| 69 | |
Guido van Rossum | 45ec02a | 2002-08-19 21:43:18 +0000 | [diff] [blame] | 70 | - A subtle change to the semantics of the built-in function intern(): |
| 71 | interned strings are no longer immortal. You must keep a reference |
| 72 | to the return value intern() around to get the benefit. |
| 73 | |
Guido van Rossum | f808b89 | 2002-08-16 03:38:10 +0000 | [diff] [blame] | 74 | - Use of 'None' as a variable, argument or attribute name now |
| 75 | issues a SyntaxWarning. In the future, None may become a keyword. |
| 76 | |
Michael W. Hudson | dd32a91 | 2002-08-15 14:59:02 +0000 | [diff] [blame] | 77 | - SET_LINENO is gone. co_lnotab is now consulted to determine when to |
| 78 | call the trace function. C code that accessed f_lineno should call |
| 79 | PyCode_Addr2Line instead (f_lineno is still there, but not kept up |
| 80 | to date). |
| 81 | |
Guido van Rossum | 9be8946 | 2002-08-14 16:11:30 +0000 | [diff] [blame] | 82 | - There's a new warning category, FutureWarning. This is used to warn |
| 83 | about a number of situations where the value or sign of an integer |
| 84 | result will change in Python 2.4 as a result of PEP 237 (integer |
| 85 | unification). The warnings implement stage B0 mentioned in that |
| 86 | PEP. The warnings are about the following situations: |
| 87 | |
| 88 | - Octal and hex literals without 'L' prefix in the inclusive range |
| 89 | [0x80000000..0xffffffff]; these are currently negative ints, but |
| 90 | in Python 2.4 they will be positive longs with the same bit |
| 91 | pattern. |
| 92 | |
| 93 | - Left shifts on integer values that cause the outcome to lose |
| 94 | bits or have a different sign than the left operand. To be |
| 95 | precise: x<<n where this currently doesn't yield the same value |
| 96 | as long(x)<<n; in Python 2.4, the outcome will be long(x)<<n. |
| 97 | |
| 98 | - Conversions from ints to string that show negative values as |
| 99 | unsigned ints in the inclusive range [0x80000000..0xffffffff]; |
| 100 | this affects the functions hex() and oct(), and the string |
| 101 | formatting codes %u, %o, %x, and %X. In Python 2.4, these will |
| 102 | show signed values (e.g. hex(-1) currently returns "0xffffffff"; |
| 103 | in Python 2.4 it will return "-0x1"). |
| 104 | |
Skip Montanaro | 979c74b | 2002-09-03 21:25:14 +0000 | [diff] [blame] | 105 | - The bits manipulated under the cover by sys.setcheckinterval() have |
| 106 | been changed. Both the check interval and the ticker used to be |
| 107 | per-thread values. They are now just a pair of global variables. In |
| 108 | addition, the default check interval was boosted from 10 to 100 |
Neal Norwitz | 7165af2 | 2002-09-05 21:42:24 +0000 | [diff] [blame] | 109 | bytecode instructions. This may have some effect on systems that |
| 110 | relied on the old default value. In particular, in multi-threaded |
| 111 | applications which try to be highly responsive, response time will |
| 112 | increase by some (perhaps imperceptible) amount. |
Skip Montanaro | 979c74b | 2002-09-03 21:25:14 +0000 | [diff] [blame] | 113 | |
Tim Peters | d64c1de | 2002-08-12 17:36:03 +0000 | [diff] [blame] | 114 | - When multiplying very large integers, a version of the so-called |
| 115 | Karatsuba algorithm is now used. This is most effective if the |
| 116 | inputs have roughly the same size. If they both have about N digits, |
| 117 | Karatsuba multiplication has O(N**1.58) runtime (the exponent is |
| 118 | log_base_2(3)) instead of the previous O(N**2). Measured results may |
Tim Peters | 012c0a3 | 2002-08-16 03:40:07 +0000 | [diff] [blame] | 119 | be better or worse than that, depending on platform quirks. Besides |
| 120 | the O() improvement in raw instruction count, the Karatsuba algorithm |
| 121 | appears to have much better cache behavior on extremely large integers |
| 122 | (starting in the ballpark of a million bits). Note that this is a |
| 123 | simple implementation, and there's no intent here to compete with, |
| 124 | e.g., GMP. It gives a very nice speedup when it applies, but a package |
| 125 | devoted to fast large-integer arithmetic should run circles around it. |
Tim Peters | 5af4e6c | 2002-08-12 02:31:19 +0000 | [diff] [blame] | 126 | |
Marc-André Lemburg | cc8764c | 2002-08-11 12:23:04 +0000 | [diff] [blame] | 127 | - u'%c' will now raise a ValueError in case the argument is an |
| 128 | integer outside the valid range of Unicode code point ordinals. |
| 129 | |
Guido van Rossum | deb0936 | 2002-08-09 17:16:30 +0000 | [diff] [blame] | 130 | - The tempfile module has been overhauled for enhanced security. The |
| 131 | mktemp() function is now deprecated; new, safe replacements are |
| 132 | mkstemp() (for files) and mkdtemp() (for directories), and the |
| 133 | higher-level functions NamedTemporaryFile() and TemporaryFile(). |
| 134 | Use of some global variables in this module is also deprecated; the |
| 135 | new functions have keyword arguments to provide the same |
| 136 | functionality. All Lib, Tools and Demo modules that used the unsafe |
| 137 | interfaces have been updated to use the safe replacements. Thanks |
| 138 | to Zack Weinberg! |
| 139 | |
Guido van Rossum | d81a983 | 2002-08-09 15:57:34 +0000 | [diff] [blame] | 140 | - When x is an object whose class implements __mul__ and __rmul__, |
| 141 | 1.0*x would correctly invoke __rmul__, but 1*x would erroneously |
| 142 | invoke __mul__. This was due to the sequence-repeat code in the int |
| 143 | type. This has been fixed now. |
| 144 | |
Barry Warsaw | d79f683 | 2002-08-06 17:01:51 +0000 | [diff] [blame] | 145 | - Previously, "str1 in str2" required str1 to be a string of length 1. |
| 146 | This restriction has been relaxed to allow str1 to be a string of |
| 147 | any length. Thus "'el' in 'hello world'" returns True now. |
| 148 | |
Guido van Rossum | b57089c | 2002-08-06 16:20:26 +0000 | [diff] [blame] | 149 | - File objects are now their own iterators. For a file f, iter(f) now |
| 150 | returns f (unless f is closed), and f.next() is similar to |
| 151 | f.readline() when EOF is not reached; however, f.next() uses a |
| 152 | readahead buffer that messes up the file position, so mixing |
| 153 | f.next() and f.readline() (or other methods) doesn't work right. |
| 154 | Calling f.seek() drops the readahead buffer, but other operations |
| 155 | don't. It so happens that this gives a nice additional speed boost |
| 156 | to "for line in file:"; the xreadlines method and corresponding |
Guido van Rossum | bffb2ef | 2002-08-09 17:17:07 +0000 | [diff] [blame] | 157 | module are now obsolete. Thanks to Oren Tirosh! |
Guido van Rossum | b57089c | 2002-08-06 16:20:26 +0000 | [diff] [blame] | 158 | |
Guido van Rossum | de392d3 | 2002-08-05 14:17:20 +0000 | [diff] [blame] | 159 | - Encoding declarations (PEP 263, phase 1) have been implemented. A |
| 160 | comment of the form "# -*- coding: <encodingname> -*-" in the first |
| 161 | or second line of a Python source file indicates the encoding. |
Martin v. Löwis | 00f1e3f | 2002-08-04 17:29:52 +0000 | [diff] [blame] | 162 | |
Tim Peters | f47630f | 2002-08-01 02:34:51 +0000 | [diff] [blame] | 163 | - list.sort() has a new implementation. While cross-platform results |
| 164 | may vary, and in data-dependent ways, this is much faster on many |
| 165 | kinds of partially ordered lists than the previous implementation, |
| 166 | and reported to be just as fast on randomly ordered lists on |
| 167 | several major platforms. This sort is also stable (if A==B and A |
| 168 | precedes B in the list at the start, A precedes B after the sort too), |
| 169 | although the language definition does not guarantee stability. A |
| 170 | potential drawback is that list.sort() may require temp space of |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 171 | len(list)*2 bytes (\*4 on a 64-bit machine). It's therefore possible |
Tim Peters | f47630f | 2002-08-01 02:34:51 +0000 | [diff] [blame] | 172 | for list.sort() to raise MemoryError now, even if a comparison function |
| 173 | does not. See <http://www.python.org/sf/587076> for full details. |
| 174 | |
Guido van Rossum | 2d5389c | 2002-07-23 03:44:35 +0000 | [diff] [blame] | 175 | - All standard iterators now ensure that, once StopIteration has been |
| 176 | raised, all future calls to next() on the same iterator will also |
| 177 | raise StopIteration. There used to be various counterexamples to |
| 178 | this behavior, which could caused confusion or subtle program |
| 179 | breakage, without any benefits. (Note that this is still an |
| 180 | iterator's responsibility; the iterator framework does not enforce |
| 181 | this.) |
| 182 | |
Mark Hammond | c0e3515 | 2002-07-16 01:32:30 +0000 | [diff] [blame] | 183 | - Ctrl+C handling on Windows has been made more consistent with |
| 184 | other platforms. KeyboardInterrupt can now reliably be caught, |
Raymond Hettinger | f2f2a2c | 2002-08-19 14:25:03 +0000 | [diff] [blame] | 185 | and Ctrl+C at an interactive prompt no longer terminates the |
Mark Hammond | c0e3515 | 2002-07-16 01:32:30 +0000 | [diff] [blame] | 186 | process under NT/2k/XP (it never did under Win9x). Ctrl+C will |
Tim Peters | 5af4e6c | 2002-08-12 02:31:19 +0000 | [diff] [blame] | 187 | interrupt time.sleep() in the main thread, and any child processes |
| 188 | created via the popen family (on win2k; we can't make win9x work |
Mark Hammond | 8f3afc7 | 2002-08-10 06:26:31 +0000 | [diff] [blame] | 189 | reliably) are also interrupted (as generally happens on for Linux/Unix.) |
| 190 | [SF bugs 231273, 439992 and 581232] |
Mark Hammond | c0e3515 | 2002-07-16 01:32:30 +0000 | [diff] [blame] | 191 | |
Raymond Hettinger | 5a04aec | 2002-06-25 00:25:30 +0000 | [diff] [blame] | 192 | - Slices and repetitions of buffer objects now consistently return |
| 193 | a string. Formerly, strings would be returned most of the time, |
| 194 | but a buffer object would be returned when the repetition count |
| 195 | was one or when the slice range was all inclusive. |
| 196 | |
Walter Dörwald | 3430d70 | 2002-06-17 10:43:59 +0000 | [diff] [blame] | 197 | - Unicode objects in sys.path are no longer ignored but treated |
| 198 | as directory names. |
| 199 | |
Neal Norwitz | 1f68fc7 | 2002-06-14 00:50:42 +0000 | [diff] [blame] | 200 | - Fixed string.startswith and string.endswith builtin methods |
| 201 | so they accept negative indices. [SF bug 493951] |
| 202 | |
Guido van Rossum | c5fe5eb | 2002-06-12 03:45:21 +0000 | [diff] [blame] | 203 | - Fixed a bug with a continue inside a try block and a yield in the |
| 204 | finally clause. [SF bug 567538] |
| 205 | |
Michael W. Hudson | 5efaf7e | 2002-06-11 10:55:12 +0000 | [diff] [blame] | 206 | - Most builtin sequences now support "extended slices", i.e. slices |
Guido van Rossum | 2e82e71 | 2002-06-13 11:41:07 +0000 | [diff] [blame] | 207 | with a third "stride" parameter. For example, "hello world"[::-1] |
| 208 | gives "dlrow olleh". |
Michael W. Hudson | 5efaf7e | 2002-06-11 10:55:12 +0000 | [diff] [blame] | 209 | |
Neal Norwitz | d68f517 | 2002-05-29 15:54:55 +0000 | [diff] [blame] | 210 | - A new warning PendingDeprecationWarning was added to provide |
| 211 | direction on features which are in the process of being deprecated. |
| 212 | The warning will not be printed by default. To see the pending |
| 213 | deprecations, use -Walways::PendingDeprecationWarning:: |
| 214 | as a command line option or warnings.filterwarnings() in code. |
| 215 | |
Fred Drake | d0ba636 | 2002-05-02 21:28:26 +0000 | [diff] [blame] | 216 | - Deprecated features of xrange objects have been removed as |
| 217 | promised. The start, stop, and step attributes and the tolist() |
| 218 | method no longer exist. xrange repetition and slicing have been |
| 219 | removed. |
| 220 | |
Guido van Rossum | 7dab242 | 2002-04-26 19:40:56 +0000 | [diff] [blame] | 221 | - New builtin function enumerate(x), from PEP 279. Example: |
| 222 | enumerate("abc") is an iterator returning (0,"a"), (1,"b"), (2,"c"). |
| 223 | The argument can be an arbitrary iterable object. |
| 224 | |
Neil Schemenauer | 89e3ee0 | 2002-04-26 01:58:53 +0000 | [diff] [blame] | 225 | - The assert statement no longer tests __debug__ at runtime. This means |
| 226 | that assert statements cannot be disabled by assigning a false value |
| 227 | to __debug__. |
| 228 | |
Walter Dörwald | 068325e | 2002-04-15 13:36:47 +0000 | [diff] [blame] | 229 | - A method zfill() was added to str and unicode, that fills a numeric |
| 230 | string to the left with zeros. For example, |
| 231 | "+123".zfill(6) -> "+00123". |
| 232 | |
Guido van Rossum | 862fe3c | 2002-04-15 12:36:47 +0000 | [diff] [blame] | 233 | - Complex numbers supported divmod() and the // and % operators, but |
| 234 | these make no sense. Since this was documented, they're being |
| 235 | deprecated now. |
| 236 | |
Guido van Rossum | 26dd830 | 2002-04-26 20:11:29 +0000 | [diff] [blame] | 237 | - String and unicode methods lstrip(), rstrip() and strip() now take |
| 238 | an optional argument that specifies the characters to strip. For |
| 239 | example, "Foo!!!?!?!?".rstrip("?!") -> "Foo". |
Guido van Rossum | 4904204 | 2002-04-13 00:59:05 +0000 | [diff] [blame] | 240 | |
Guido van Rossum | c69343b | 2002-04-12 19:22:48 +0000 | [diff] [blame] | 241 | - Added a new dict method pop(key). This removes and returns the |
| 242 | value corresponding to key. [SF patch #539949] |
| 243 | |
Guido van Rossum | 4783446 | 2002-04-04 15:21:33 +0000 | [diff] [blame] | 244 | - A new built-in type, bool, has been added, as well as built-in |
| 245 | names for its two values, True and False. Comparisons and sundry |
| 246 | other operations that return a truth value have been changed to |
Andrew M. Kuchling | 21b23b0 | 2002-05-07 20:58:03 +0000 | [diff] [blame] | 247 | return a bool instead. Read PEP 285 for an explanation of why this |
Guido van Rossum | 4783446 | 2002-04-04 15:21:33 +0000 | [diff] [blame] | 248 | is backward compatible. |
| 249 | |
Guido van Rossum | 95b62a6 | 2002-03-29 01:07:24 +0000 | [diff] [blame] | 250 | - Fixed two bugs reported as SF #535905: under certain conditions, |
| 251 | deallocating a deeply nested structure could cause a segfault in the |
| 252 | garbage collector, due to interaction with the "trashcan" code; |
| 253 | access to the current frame during destruction of a local variable |
| 254 | could access a pointer to freed memory. |
| 255 | |
Guido van Rossum | fe0ca4a | 2002-06-18 21:20:13 +0000 | [diff] [blame] | 256 | - The optional object allocator ("pymalloc") has been enabled by |
| 257 | default. The recommended practice for memory allocation and |
| 258 | deallocation has been streamlined. A header file is included, |
| 259 | Misc/pymemcompat.h, which can be bundled with 3rd party extensions |
| 260 | and lets them use the same API with Python versions from 1.5.2 |
| 261 | onwards. |
Neil Schemenauer | 9d50d13 | 2002-03-22 17:06:59 +0000 | [diff] [blame] | 262 | |
Martin v. Löwis | cfeb3b6 | 2002-03-03 21:30:27 +0000 | [diff] [blame] | 263 | - PyErr_Display will provide file and line information for all exceptions |
| 264 | that have an attribute print_file_and_line, not just SyntaxErrors. |
| 265 | |
Marc-André Lemburg | e92e0a9 | 2002-02-09 11:28:43 +0000 | [diff] [blame] | 266 | - The UTF-8 codec will now encode and decode Unicode surrogates |
| 267 | correctly and without raising exceptions for unpaired ones. |
| 268 | |
Guido van Rossum | add8806 | 2002-08-15 14:01:14 +0000 | [diff] [blame] | 269 | - Universal newlines (PEP 278) is implemented. Briefly, using 'U' |
| 270 | instead of 'r' when opening a text file for reading changes the line |
| 271 | ending convention so that any of '\r', '\r\n', and '\n' is |
| 272 | recognized (even mixed in one file); all three are converted to |
| 273 | '\n', the standard Python line end character. |
| 274 | |
Neal Norwitz | 649b759 | 2002-01-01 19:07:13 +0000 | [diff] [blame] | 275 | - file.xreadlines() now raises a ValueError if the file is closed: |
| 276 | Previously, an xreadlines object was returned which would raise |
| 277 | a ValueError when the xreadlines.next() method was called. |
| 278 | |
Andrew M. Kuchling | 21b23b0 | 2002-05-07 20:58:03 +0000 | [diff] [blame] | 279 | - sys.exit() inadvertently allowed more than one argument. |
Neal Norwitz | 0c766a0 | 2002-03-27 13:03:09 +0000 | [diff] [blame] | 280 | An exception will now be raised if more than one argument is used. |
| 281 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 282 | Extension modules |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 283 | ----------------- |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 284 | |
Martin v. Löwis | 658009a | 2002-09-16 17:26:24 +0000 | [diff] [blame] | 285 | - cPickle.BadPickleGet is now a class. |
| 286 | |
Martin v. Löwis | a32c994 | 2002-09-09 16:17:47 +0000 | [diff] [blame] | 287 | - The time stamps in os.stat_result are floating point numbers now. |
Martin v. Löwis | 94717ed | 2002-09-09 14:24:16 +0000 | [diff] [blame] | 288 | |
Neal Norwitz | b567392 | 2002-09-05 21:48:07 +0000 | [diff] [blame] | 289 | - If the size passed to mmap.mmap() is larger than the length of the |
| 290 | file on non-Windows platforms, a ValueError is raised. [SF bug 585792] |
| 291 | |
Guido van Rossum | b57089c | 2002-08-06 16:20:26 +0000 | [diff] [blame] | 292 | - The xreadlines module is slated for obsolescence. |
| 293 | |
Guido van Rossum | 79f0a10 | 2002-07-23 03:32:08 +0000 | [diff] [blame] | 294 | - The strptime function in the time module is now always available (a |
| 295 | Python implementation is used when the C library doesn't define it). |
| 296 | |
Guido van Rossum | 9562bcf | 2002-06-14 21:31:18 +0000 | [diff] [blame] | 297 | - The 'new' module is no longer an extension, but a Python module that |
| 298 | only exists for backwards compatibility. Its contents are no longer |
| 299 | functions but callable type objects. |
| 300 | |
Tim Peters | 81b9251 | 2002-04-29 01:37:32 +0000 | [diff] [blame] | 301 | - The bsddb.*open functions can now take 'None' as a filename. |
Anthony Baxter | 8388895 | 2002-04-23 02:11:05 +0000 | [diff] [blame] | 302 | This will create a temporary in-memory bsddb that won't be |
Tim Peters | 81b9251 | 2002-04-29 01:37:32 +0000 | [diff] [blame] | 303 | written to disk. |
Anthony Baxter | 8388895 | 2002-04-23 02:11:05 +0000 | [diff] [blame] | 304 | |
Martin v. Löwis | 0cec0ff | 2002-07-28 16:33:45 +0000 | [diff] [blame] | 305 | - posix.lchown, posix.killpg, posix.mknod, and posix.getpgid have been |
| 306 | added where available. |
Martin v. Löwis | 06a83e9 | 2002-04-14 10:19:44 +0000 | [diff] [blame] | 307 | |
Martin v. Löwis | 2e64c34 | 2002-03-27 18:49:02 +0000 | [diff] [blame] | 308 | - The locale module now exposes the C library's gettext interface. |
| 309 | |
Tim Peters | fbf74f2 | 2002-03-14 19:06:01 +0000 | [diff] [blame] | 310 | - A security hole ("double free") was found in zlib-1.1.3, a popular |
| 311 | third party compression library used by some Python modules. The |
| 312 | hole was quickly plugged in zlib-1.1.4, and the Windows build of |
| 313 | Python now ships with zlib-1.1.4. |
| 314 | |
Martin v. Löwis | 688357e | 2002-04-08 21:28:20 +0000 | [diff] [blame] | 315 | - pwd, grp, and resource return enhanced tuples now, with symbolic |
| 316 | field names. |
Martin v. Löwis | 29001ff | 2002-03-01 10:47:37 +0000 | [diff] [blame] | 317 | |
Martin v. Löwis | 9986633 | 2002-03-01 10:27:01 +0000 | [diff] [blame] | 318 | - array.array is now a type object. A new format character |
| 319 | 'u' indicates Py_UNICODE arrays. For those, .tounicode and |
| 320 | .fromunicode methods are available. Arrays now support __iadd__ |
| 321 | and __imul__. |
| 322 | |
Neal Norwitz | e790eae | 2002-01-01 21:14:12 +0000 | [diff] [blame] | 323 | - dl now builds on every system that has dlfcn.h. Failure in case |
Martin v. Löwis | 9322727 | 2002-01-01 20:18:30 +0000 | [diff] [blame] | 324 | of sizeof(int)!=sizeof(long)!=sizeof(void*) is delayed until dl.open |
| 325 | is called. |
| 326 | |
Michael W. Hudson | 34f20ea | 2002-05-27 15:08:24 +0000 | [diff] [blame] | 327 | - signal.sigpending, signal.sigprocmask and signal.sigsuspend have |
| 328 | been added where available. |
| 329 | |
Skip Montanaro | 8e790e7 | 2002-09-03 13:25:17 +0000 | [diff] [blame] | 330 | - The sys module acquired a new attribute, api_version, which evaluates |
| 331 | to the value of the PYTHON_API_VERSION macro with which the |
| 332 | interpreter was compiled. |
| 333 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 334 | Library |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 335 | ------- |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 336 | |
Walter Dörwald | 5ccaf8f | 2002-09-06 16:15:58 +0000 | [diff] [blame] | 337 | - mimetypes has two new functions: guess_all_extensions() which |
| 338 | returns a list of all known extensions for a mime type, and |
| 339 | add_type() which adds one mapping between a mime type and |
| 340 | an extension to the database. |
| 341 | |
Guido van Rossum | d8ab35c | 2002-08-19 16:25:46 +0000 | [diff] [blame] | 342 | - New module: sets, defines the class Set that implements a mutable |
| 343 | set type using the keys of a dict to represent the set. There's |
| 344 | also a class ImmutableSet which is useful when you need sets of sets |
| 345 | or when you need to use sets as dict keys, and a class BaseSet which |
| 346 | is the base class of the two. (This is not documented yet, but |
| 347 | help(sets) gives a wealth of information.) |
| 348 | |
Raymond Hettinger | f2f2a2c | 2002-08-19 14:25:03 +0000 | [diff] [blame] | 349 | - Added operator.pow(a,b) which is equivalent to a**b. |
| 350 | |
Tim Peters | 012c0a3 | 2002-08-16 03:40:07 +0000 | [diff] [blame] | 351 | - random.randrange(-sys.maxint-1, sys.maxint) no longer raises |
| 352 | OverflowError. That is, it now accepts any combination of 'start' |
| 353 | and 'stop' arguments so long as each is in the range of Python's |
| 354 | bounded integers. |
| 355 | |
Guido van Rossum | 4a57c33 | 2002-08-02 18:05:20 +0000 | [diff] [blame] | 356 | - New "algorithms" module: heapq, implements a heap queue. Thanks to |
| 357 | Kevin O'Connor for the code and François Pinard for an entertaining |
| 358 | write-up explaining the theory and practical uses of heaps. |
| 359 | |
Marc-André Lemburg | 58e2ff2 | 2002-07-12 14:40:04 +0000 | [diff] [blame] | 360 | - New encoding for the Palm OS character set: palmos. |
| 361 | |
Tim Peters | 943382c | 2002-07-07 03:59:34 +0000 | [diff] [blame] | 362 | - binascii.crc32() and the zipfile module had problems on some 64-bit |
| 363 | platforms. These have been fixed. On a platform with 8-byte C longs, |
| 364 | crc32() now returns a signed-extended 4-byte result, so that its value |
| 365 | as a Python int is equal to the value computed a 32-bit platform. |
| 366 | |
Martin v. Löwis | 7d650ca | 2002-06-30 15:05:00 +0000 | [diff] [blame] | 367 | - xml.dom.minidom.toxml and toprettyxml now take an optional encoding |
| 368 | argument. |
| 369 | |
Guido van Rossum | 9f7549b | 2002-06-10 21:39:42 +0000 | [diff] [blame] | 370 | - Some fixes in the copy module: when an object is copied through its |
| 371 | __reduce__ method, there was no check for a __setstate__ method on |
| 372 | the result [SF patch 565085]; deepcopy should treat instances of |
| 373 | custom metaclasses the same way it treats instances of type 'type' |
| 374 | [SF patch 560794]. |
| 375 | |
Guido van Rossum | 5033c8e | 2002-06-07 12:40:52 +0000 | [diff] [blame] | 376 | - Sockets now support timeout mode. After s.settimeout(T), where T is |
| 377 | a float expressing seconds, subsequent operations raise an exception |
| 378 | if they cannot be completed within T seconds. To disable timeout |
Guido van Rossum | 79f0a10 | 2002-07-23 03:32:08 +0000 | [diff] [blame] | 379 | mode, use s.settimeout(None). There's also a module function, |
| 380 | socket.setdefaulttimeout(T), which sets the default for all sockets |
| 381 | created henceforth. |
Guido van Rossum | 5033c8e | 2002-06-07 12:40:52 +0000 | [diff] [blame] | 382 | |
| 383 | - getopt.gnu_getopt was added. This supports GNU-style option |
| 384 | processing, where options can be mixed with non-option arguments. |
Martin v. Löwis | 446a25f | 2002-06-06 10:58:36 +0000 | [diff] [blame] | 385 | |
Michael W. Hudson | b0dc1a3 | 2002-06-04 18:27:35 +0000 | [diff] [blame] | 386 | - Stop using strings for exceptions. String objects used for |
| 387 | exceptions are now classes deriving from Exception. The objects |
| 388 | changed were: Tkinter.TclError, bdb.BdbQuit, macpath.norm_error, |
| 389 | tabnanny.NannyNag, and xdrlib.Error. |
Neal Norwitz | b5b5a26 | 2002-06-04 17:14:07 +0000 | [diff] [blame] | 390 | |
Walter Dörwald | 474458d | 2002-06-04 15:16:29 +0000 | [diff] [blame] | 391 | - Constants BOM_UTF8, BOM_UTF16, BOM_UTF16_LE, BOM_UTF16_BE, |
| 392 | BOM_UTF32, BOM_UTF32_LE and BOM_UTF32_BE that represent the Byte |
| 393 | Order Mark in UTF-8, UTF-16 and UTF-32 encodings for little and |
| 394 | big endian systems were added to the codecs module. The old names |
| 395 | BOM32_* and BOM64_* were off by a factor of 2. |
| 396 | |
Guido van Rossum | f925782 | 2002-06-18 20:43:18 +0000 | [diff] [blame] | 397 | - Added conversion functions math.degrees() and math.radians(). |
Raymond Hettinger | 87f59ee | 2002-05-13 04:17:32 +0000 | [diff] [blame] | 398 | |
Raymond Hettinger | 9b1df1d | 2002-05-12 06:07:21 +0000 | [diff] [blame] | 399 | - ftplib.retrlines() now tests for callback is None rather than testing |
| 400 | for False. Was causing an error when given a callback object which |
| 401 | was callable but also returned len() as zero. The change may |
| 402 | create new breakage if the caller relied on the undocumented behavior |
| 403 | and called with callback set to [] or some other False value not |
| 404 | identical to None. |
| 405 | |
Tim Peters | 46c04e1 | 2002-05-05 20:40:00 +0000 | [diff] [blame] | 406 | - random.gauss() uses a piece of hidden state used by nothing else, |
| 407 | and the .seed() and .whseed() methods failed to reset it. In other |
| 408 | words, setting the seed didn't completely determine the sequence of |
| 409 | results produced by random.gauss(). It does now. Programs repeatedly |
| 410 | mixing calls to a seed method with calls to gauss() may see different |
| 411 | results now. |
| 412 | |
Fred Drake | 7f781c9 | 2002-05-01 20:33:53 +0000 | [diff] [blame] | 413 | - The pickle.Pickler class grew a clear_memo() method to mimic that |
| 414 | provided by cPickle.Pickler. |
| 415 | |
Tim Peters | 81b9251 | 2002-04-29 01:37:32 +0000 | [diff] [blame] | 416 | - difflib's SequenceMatcher class now does a dynamic analysis of |
| 417 | which elements are so frequent as to constitute noise. For |
| 418 | comparing files as sequences of lines, this generally works better |
| 419 | than the IS_LINE_JUNK function, and function ndiff's linejunk |
| 420 | argument defaults to None now as a result. A happy benefit is |
| 421 | that SequenceMatcher may run much faster now when applied |
| 422 | to large files with many duplicate lines (for example, C program |
| 423 | text with lots of repeated "}" and "return NULL;" lines). |
| 424 | |
Guido van Rossum | 256705b | 2002-04-23 13:29:43 +0000 | [diff] [blame] | 425 | - New Text.dump() method in Tkinter module. |
| 426 | |
Marc-André Lemburg | 1ec71ea | 2002-04-17 20:33:40 +0000 | [diff] [blame] | 427 | - New distutils commands for building packagers were added to |
| 428 | support pkgtool on Solaris and swinstall on HP-UX. |
| 429 | |
| 430 | - distutils now has a new abstract binary packager base class |
| 431 | command/bdist_packager, which simplifies writing packagers. |
| 432 | This will hopefully provide the missing bits to encourage |
| 433 | people to submit more packagers, e.g. for Debian, FreeBSD |
| 434 | and other systems. |
| 435 | |
Guido van Rossum | f925782 | 2002-06-18 20:43:18 +0000 | [diff] [blame] | 436 | - The UTF-16, -LE and -BE stream readers now raise a |
| 437 | NotImplementedError for all calls to .readline(). Previously, they |
| 438 | used to just produce garbage or fail with an encoding error -- |
| 439 | UTF-16 is a 2-byte encoding and the C lib's line reading APIs don't |
Marc-André Lemburg | 4179a01 | 2002-04-05 12:15:05 +0000 | [diff] [blame] | 440 | work well with these. |
| 441 | |
Martin v. Löwis | 5c137c2 | 2002-03-18 12:44:08 +0000 | [diff] [blame] | 442 | - compileall now supports quiet operation. |
| 443 | |
Andrew M. Kuchling | 21b23b0 | 2002-05-07 20:58:03 +0000 | [diff] [blame] | 444 | - The BaseHTTPServer now implements optional HTTP/1.1 persistent |
Martin v. Löwis | 587c98c | 2002-03-17 18:37:22 +0000 | [diff] [blame] | 445 | connections. |
| 446 | |
Marc-André Lemburg | a5d2b4c | 2002-02-16 18:23:30 +0000 | [diff] [blame] | 447 | - socket module: the SSL support was broken out of the main |
| 448 | _socket module C helper and placed into a new _ssl helper |
| 449 | which now gets imported by socket.py if available and working. |
| 450 | |
Marc-André Lemburg | 37aa580 | 2002-02-10 21:42:47 +0000 | [diff] [blame] | 451 | - encodings package: added aliases for all supported IANA character |
| 452 | sets |
| 453 | |
Guido van Rossum | 606fe4e | 2001-12-28 20:57:14 +0000 | [diff] [blame] | 454 | - ftplib: to safeguard the user's privacy, anonymous login will use |
| 455 | "anonymous@" as default password, rather than the real user and host |
| 456 | name. |
| 457 | |
Fred Drake | 3673612 | 2002-01-10 13:50:31 +0000 | [diff] [blame] | 458 | - webbrowser: tightened up the command passed to os.system() so that |
| 459 | arbitrary shell code can't be executed because a bogus URL was |
| 460 | passed in. |
| 461 | |
Tim Peters | 11bd994 | 2002-02-01 00:52:29 +0000 | [diff] [blame] | 462 | - gettext.translation has an optional fallback argument, and |
Martin v. Löwis | a55ffae | 2002-01-11 06:58:49 +0000 | [diff] [blame] | 463 | gettext.find an optional all argument. Translations will now fallback |
| 464 | on a per-message basis. |
Martin v. Löwis | 1be6419 | 2002-01-11 06:33:28 +0000 | [diff] [blame] | 465 | |
Martin v. Löwis | 9668b93 | 2002-01-12 11:27:42 +0000 | [diff] [blame] | 466 | - distutils bdist commands now offer a --skip-build option. |
| 467 | |
Walter Dörwald | b25c2b0 | 2002-03-21 10:38:40 +0000 | [diff] [blame] | 468 | - warnings.warn now accepts a Warning instance as first argument. |
| 469 | |
Fred Drake | fed7e85 | 2002-04-04 19:36:15 +0000 | [diff] [blame] | 470 | - The xml.sax.expatreader.ExpatParser class will no longer create |
| 471 | circular references by using itself as the locator that gets passed |
| 472 | to the content handler implementation. [SF bug #535474] |
| 473 | |
Barry Warsaw | 47f3e2c | 2002-05-20 00:14:24 +0000 | [diff] [blame] | 474 | - The email.Parser.Parser class now properly parses strings regardless |
| 475 | of their line endings, which can be any of \r, \n, or \r\n (CR, LF, |
| 476 | or CRLF). Also, the Header class's constructor default arguments |
| 477 | has changed slightly so that an explicit maxlinelen value is always |
| 478 | honored. |
| 479 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 480 | Tools/Demos |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 481 | ----------- |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 482 | |
Guido van Rossum | 923e4ef | 2002-09-17 20:58:59 +0000 | [diff] [blame] | 483 | - The SGI demos (Demo/sgi) have been removed. Nobody thought they |
| 484 | were interesting any more. (The SGI library modules and extensions |
| 485 | are still there; it is believed that at least some of these are |
| 486 | still used and useful.) |
| 487 | |
Fred Drake | 40813b1 | 2002-08-05 15:24:19 +0000 | [diff] [blame] | 488 | - IDLE supports the new encoding declarations (PEP 263); it can also |
| 489 | deal with legacy 8-bit files if they use the locale's encoding. It |
| 490 | allows non-ASCII strings in the interactive shell and executes them |
| 491 | in the locale's encoding. |
Martin v. Löwis | 98ce7b7 | 2002-08-05 15:11:26 +0000 | [diff] [blame] | 492 | |
Marc-André Lemburg | 4626009 | 2002-04-04 16:17:11 +0000 | [diff] [blame] | 493 | - freeze.py now produces binaries which can import shared modules, |
| 494 | unlike before when this failed due to missing symbol exports in |
| 495 | the generated binary. |
| 496 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 497 | Build |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 498 | ----- |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 499 | |
Tim Peters | f47630f | 2002-08-01 02:34:51 +0000 | [diff] [blame] | 500 | - The public Python C API will generally be declared using PyAPI_FUNC |
Neal Norwitz | 1389f3e | 2002-07-22 13:21:10 +0000 | [diff] [blame] | 501 | and PyAPI_DATA macros, while Python extension module init functions |
| 502 | will be declared with PyMODINIT_FUNC. DL_EXPORT/DL_IMPORT macros |
| 503 | are deprecated. |
| 504 | |
Tim Peters | c6a3ff6 | 2002-07-08 22:11:52 +0000 | [diff] [blame] | 505 | - A bug was fixed that could cause COUNT_ALLOCS builds to segfault, or |
| 506 | get into infinite loops, when a new-style class got garbage-collected. |
| 507 | Unfortunately, to avoid this, the way COUNT_ALLOCS works requires |
| 508 | that new-style classes be immortal in COUNT_ALLOCS builds. Note that |
| 509 | COUNT_ALLOCS is not enabled by default, in either release or debug |
| 510 | builds, and that new-style classes are immortal only in COUNT_ALLOCS |
| 511 | builds. |
| 512 | |
Tim Peters | 943382c | 2002-07-07 03:59:34 +0000 | [diff] [blame] | 513 | - Compiling out the cyclic garbage collector is no longer an option. |
| 514 | The old symbol WITH_CYCLE_GC is now ignored, and Python.h arranges |
| 515 | that it's always defined (for the benefit of any extension modules |
| 516 | that may be conditionalizing on it). A bonus is that any extension |
| 517 | type participating in cyclic gc can choose to participate in the |
| 518 | Py_TRASHCAN mechanism now too; in the absence of cyclic gc, this used |
| 519 | to require editing the core to teach the trashcan mechanism about the |
| 520 | new type. |
| 521 | |
Martin v. Löwis | 0cec0ff | 2002-07-28 16:33:45 +0000 | [diff] [blame] | 522 | - According to Annex F of the current C standard, |
Tim Peters | 1de41bf | 2002-07-03 03:31:20 +0000 | [diff] [blame] | 523 | |
| 524 | The Standard C macro HUGE_VAL and its float and long double analogs, |
| 525 | HUGE_VALF and HUGE_VALL, expand to expressions whose values are |
| 526 | positive infinities. |
| 527 | |
| 528 | Python only uses the double HUGE_VAL, and only to #define its own symbol |
| 529 | Py_HUGE_VAL. Some platforms have incorrect definitions for HUGE_VAL. |
| 530 | pyport.h used to try to worm around that, but the workarounds triggered |
| 531 | other bugs on other platforms, so we gave up. If your platform defines |
| 532 | HUGE_VAL incorrectly, you'll need to #define Py_HUGE_VAL to something |
| 533 | that works on your platform. The only instance of this I'm sure about |
| 534 | is on an unknown subset of Cray systems, described here: |
| 535 | |
| 536 | http://www.cray.com/swpubs/manuals/SN-2194_2.0/html-SN-2194_2.0/x3138.htm |
| 537 | |
| 538 | Presumably 2.3a1 breaks such systems. If anyone uses such a system, help! |
| 539 | |
Martin v. Löwis | a3fb4f7 | 2002-06-09 13:33:54 +0000 | [diff] [blame] | 540 | - The configure option --without-doc-strings can be used to remove the |
| 541 | doc strings from the builtin functions and modules; this reduces the |
| 542 | size of the executable. |
| 543 | |
Guido van Rossum | add8806 | 2002-08-15 14:01:14 +0000 | [diff] [blame] | 544 | - The universal newlines option (PEP 278) is on by default. On Unix |
| 545 | it can be disabled by passing --without-universal-newlines to the |
| 546 | configure script. On other platforms, remove |
| 547 | WITH_UNIVERSAL_NEWLINES from pyconfig.h. |
Tim Peters | 639295f | 2002-04-21 07:30:30 +0000 | [diff] [blame] | 548 | |
Martin v. Löwis | 1142de3 | 2002-03-29 16:28:31 +0000 | [diff] [blame] | 549 | - On Unix, a shared libpython2.3.so can be created with --enable-shared. |
| 550 | |
Tim Peters | 8deda70 | 2002-03-30 10:06:07 +0000 | [diff] [blame] | 551 | - All uses of the CACHE_HASH, INTERN_STRINGS, and DONT_SHARE_SHORT_STRINGS |
| 552 | preprocessor symbols were eliminated. The internal decisions they |
| 553 | controlled stopped being experimental long ago. |
Tim Peters | 1f7df35 | 2002-03-29 03:29:08 +0000 | [diff] [blame] | 554 | |
Fred Drake | 7d28b73 | 2002-05-02 21:17:00 +0000 | [diff] [blame] | 555 | - The tools used to build the documentation now work under Cygwin as |
| 556 | well as Unix. |
| 557 | |
Skip Montanaro | 57454e5 | 2002-06-14 20:30:31 +0000 | [diff] [blame] | 558 | - The bsddb and dbm module builds have been changed to try and avoid version |
| 559 | skew problems and disable linkage with Berkeley DB 1.85 unless the |
| 560 | installer knows what s/he's doing. See the section on building these |
| 561 | modules in the README file for details. |
| 562 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 563 | C API |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 564 | ----- |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 565 | |
Guido van Rossum | 45ec02a | 2002-08-19 21:43:18 +0000 | [diff] [blame] | 566 | - The string object's layout has changed: the pointer member |
| 567 | ob_sinterned has been replaced by an int member ob_sstate. On some |
| 568 | platforms (e.g. most 64-bit systems) this may change the offset of |
| 569 | the ob_sval member, so as a precaution the API_VERSION has been |
| 570 | incremented. The apparently unused feature of "indirect interned |
| 571 | strings", supported by the ob_sinterned member, is gone. Interned |
| 572 | strings are now usually mortal; theres a new API, |
| 573 | PyString_InternImmortal() that creates immortal interned strings. |
| 574 | (The ob_sstate member can only take three values; however, while |
| 575 | making it a char saves a few bytes per string object on average, in |
| 576 | it also slowed things down a bit because ob_sval was no longer |
| 577 | aligned.) |
| 578 | |
Guido van Rossum | fdb8648 | 2002-08-14 21:20:32 +0000 | [diff] [blame] | 579 | - The Py_InitModule*() functions now accept NULL for the 'methods' |
| 580 | argument. Modules without global functions are becoming more common |
| 581 | now that factories can be types rather than functions. |
| 582 | |
Marc-André Lemburg | cc8764c | 2002-08-11 12:23:04 +0000 | [diff] [blame] | 583 | - New C API PyUnicode_FromOrdinal() which exposes unichr() at C |
| 584 | level. |
| 585 | |
Thomas Heller | 085358a | 2002-07-29 14:27:41 +0000 | [diff] [blame] | 586 | - New functions PyErr_SetExcFromWindowsErr() and |
| 587 | PyErr_SetExcFromWindowsErrWithFilename(). Similar to |
| 588 | PyErr_SetFromWindowsErrWithFilename() and |
| 589 | PyErr_SetFromWindowsErr(), but they allow to specify |
| 590 | the exception type to raise. Available on Windows. |
| 591 | |
Tim Peters | 7c321a8 | 2002-07-09 02:57:01 +0000 | [diff] [blame] | 592 | - Py_FatalError() is now declared as taking a const char* argument. It |
| 593 | was previously declared without const. This should not affect working |
| 594 | code. |
| 595 | |
Martin v. Löwis | 01f94bd | 2002-05-08 08:44:21 +0000 | [diff] [blame] | 596 | - Added new macro PySequence_ITEM(o, i) that directly calls |
| 597 | sq_item without rechecking that o is a sequence and without |
| 598 | adjusting for negative indices. |
| 599 | |
Fred Drake | d0ba636 | 2002-05-02 21:28:26 +0000 | [diff] [blame] | 600 | - PyRange_New() now raises ValueError if the fourth argument is not 1. |
| 601 | This is part of the removal of deprecated features of the xrange |
| 602 | object. |
| 603 | |
Guido van Rossum | 517c7d4 | 2002-04-26 02:49:14 +0000 | [diff] [blame] | 604 | - PyNumber_Coerce() and PyNumber_CoerceEx() now also invoke the type's |
| 605 | coercion if both arguments have the same type but this type has the |
| 606 | CHECKTYPES flag set. This is to better support proxies. |
| 607 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 608 | - The type of tp_free has been changed from "void (*)(PyObject \*)" to |
| 609 | "void (*)(void \*)". |
Neil Schemenauer | 9344b14 | 2002-04-12 23:00:08 +0000 | [diff] [blame] | 610 | |
| 611 | - PyObject_Del, PyObject_GC_Del are now functions instead of macros. |
| 612 | |
Guido van Rossum | 0986d82 | 2002-04-08 01:38:42 +0000 | [diff] [blame] | 613 | - A type can now inherit its metatype from its base type. Previously, |
| 614 | when PyType_Ready() was called, if ob_type was found to be NULL, it |
| 615 | was always set to &PyType_Type; now it is set to base->ob_type, |
| 616 | where base is tp_base, defaulting to &PyObject_Type. |
| 617 | |
| 618 | - PyType_Ready() accidentally did not inherit tp_is_gc; now it does. |
| 619 | |
Neil Schemenauer | c9abc1d | 2002-05-08 14:14:41 +0000 | [diff] [blame] | 620 | - The PyCore_* family of APIs have been removed. |
Neil Schemenauer | 9d50d13 | 2002-03-22 17:06:59 +0000 | [diff] [blame] | 621 | |
Andrew M. Kuchling | 21b23b0 | 2002-05-07 20:58:03 +0000 | [diff] [blame] | 622 | - The "u#" parser marker will now pass through Unicode objects as-is |
Marc-André Lemburg | 3e3eacb | 2002-01-09 16:21:27 +0000 | [diff] [blame] | 623 | without going through the buffer API. |
| 624 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 625 | - The enumerators of cmp_op have been renamed to use the prefix ``PyCmp_``. |
Martin v. Löwis | 7198a52 | 2002-01-01 19:59:11 +0000 | [diff] [blame] | 626 | |
Tim Peters | 10a3bb5 | 2001-12-25 19:07:38 +0000 | [diff] [blame] | 627 | - An old #define of ANY as void has been removed from pyport.h. This |
| 628 | hasn't been used since Python's pre-ANSI days, and the #define has |
| 629 | been marked as obsolete since then. SF bug 495548 says it created |
| 630 | conflicts with other packages, so keeping it around wasn't harmless. |
| 631 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 632 | - Because Python's magic number scheme broke on January 1st, we decided |
| 633 | to stop Python development. Thanks for all the fish! |
| 634 | |
Tim Peters | fb05db2 | 2002-03-11 00:24:00 +0000 | [diff] [blame] | 635 | - Some of us don't like fish, so we changed Python's magic number |
Marc-André Lemburg | e92e0a9 | 2002-02-09 11:28:43 +0000 | [diff] [blame] | 636 | scheme to a new one. See Python/import.c for details. |
| 637 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 638 | New platforms |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 639 | ------------- |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 640 | |
Martin v. Löwis | f90ae20 | 2002-06-11 06:22:31 +0000 | [diff] [blame] | 641 | - AtheOS is now supported. |
| 642 | |
Andrew MacIntyre | 161ad0e | 2002-08-04 07:25:58 +0000 | [diff] [blame] | 643 | - the EMX runtime environment on OS/2 is now supported. |
| 644 | |
Martin v. Löwis | a6e9758 | 2002-01-01 18:41:33 +0000 | [diff] [blame] | 645 | - GNU/Hurd is now supported. |
| 646 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 647 | Tests |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 648 | ----- |
| 649 | |
| 650 | Yet to be written. |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 651 | |
| 652 | Windows |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 653 | ------- |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 654 | |
Tim Peters | 08e7295 | 2002-04-16 20:48:01 +0000 | [diff] [blame] | 655 | - Sometimes the uninstall executable (UNWISE.EXE) vanishes. One cause |
| 656 | of that has been fixed in the installer (disabled Wise's "delete in- |
| 657 | use files" uninstall option). |
| 658 | |
Guido van Rossum | 181e41a | 2002-04-06 00:23:25 +0000 | [diff] [blame] | 659 | - Fixed a bug in urllib's proxy handling in Windows. [SF bug #503031] |
| 660 | |
Tim Peters | 3318792 | 2002-04-04 20:02:04 +0000 | [diff] [blame] | 661 | - The installer now installs Start menu shortcuts under (the local |
| 662 | equivalent of) "All Users" when doing an Admin install. |
| 663 | |
Tim Peters | fb05db2 | 2002-03-11 00:24:00 +0000 | [diff] [blame] | 664 | - file.truncate([newsize]) now works on Windows for all newsize values. |
| 665 | It used to fail if newsize didn't fit in 32 bits, reflecting a |
| 666 | limitation of MS _chsize (which is no longer used). |
| 667 | |
Tim Peters | ab034fa | 2002-02-01 11:27:43 +0000 | [diff] [blame] | 668 | - os.waitpid() is now implemented for Windows, and can be used to block |
| 669 | until a specified process exits. This is similar to, but not exactly |
| 670 | the same as, os.waitpid() on POSIX systems. If you're waiting for |
| 671 | a specific process whose pid was obtained from one of the spawn() |
| 672 | functions, the same Python os.waitpid() code works across platforms. |
Tim Peters | 3318792 | 2002-04-04 20:02:04 +0000 | [diff] [blame] | 673 | See the docs for details. The docs were changed to clarify that |
| 674 | spawn functions return, and waitpid requires, a process handle on |
| 675 | Windows (not the same thing as a Windows process id). |
Tim Peters | ab034fa | 2002-02-01 11:27:43 +0000 | [diff] [blame] | 676 | |
Tim Peters | 11bd994 | 2002-02-01 00:52:29 +0000 | [diff] [blame] | 677 | - New tempfile.TemporaryFile implementation for Windows: this doesn't |
Andrew M. Kuchling | 21b23b0 | 2002-05-07 20:58:03 +0000 | [diff] [blame] | 678 | need a TemporaryFileWrapper wrapper anymore, and should be immune |
Tim Peters | 11bd994 | 2002-02-01 00:52:29 +0000 | [diff] [blame] | 679 | to a nasty problem: before 2.3, if you got a temp file on Windows, it |
| 680 | got wrapped in an object whose close() method first closed the |
| 681 | underlying file, then deleted the file. This usually worked fine. |
| 682 | However, the spawn family of functions on Windows create (at a low C |
| 683 | level) the same set of open files in the spawned process Q as were |
| 684 | open in the spawning process P. If a temp file f was among them, then |
| 685 | doing f.close() in P first closed P's C-level file handle on f, but Q's |
| 686 | C-level file handle on f remained open, so the attempt in P to delete f |
| 687 | blew up with a "Permission denied" error (Windows doesn't allow |
| 688 | deleting open files). This was surprising, subtle, and difficult to |
| 689 | work around. |
| 690 | |
| 691 | - The os module now exports all the symbolic constants usable with the |
| 692 | low-level os.open() on Windows: the new constants in 2.3 are |
| 693 | O_NOINHERIT, O_SHORT_LIVED, O_TEMPORARY, O_RANDOM and O_SEQUENTIAL. |
| 694 | The others were also available in 2.2: O_APPEND, O_BINARY, O_CREAT, |
| 695 | O_EXCL, O_RDONLY, O_RDWR, O_TEXT, O_TRUNC and O_WRONLY. Contrary |
| 696 | to Microsoft docs, O_SHORT_LIVED does not seem to imply O_TEMPORARY |
| 697 | (so specify both if you want both; note that neither is useful unless |
| 698 | specified with O_CREAT too). |
| 699 | |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 700 | Mac |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 701 | ---- |
| 702 | |
| 703 | Yet to be written. |
Tim Peters | af8446f | 2001-12-21 21:36:50 +0000 | [diff] [blame] | 704 | |
| 705 | |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 706 | What's New in Python 2.2 final? |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 707 | =============================== |
| 708 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 709 | *Release date: 21-Dec-2001* |
| 710 | |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 711 | Type/class unification and new-style classes |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 712 | -------------------------------------------- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 713 | |
Barry Warsaw | 52acb49 | 2001-12-21 20:04:22 +0000 | [diff] [blame] | 714 | - pickle.py, cPickle: allow pickling instances of new-style classes |
| 715 | with a custom metaclass. |
| 716 | |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 717 | Core and builtins |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 718 | ----------------- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 719 | |
Barry Warsaw | 52acb49 | 2001-12-21 20:04:22 +0000 | [diff] [blame] | 720 | - weakref proxy object: when comparing, unwrap both arguments if both |
| 721 | are proxies. |
| 722 | |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 723 | Extension modules |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 724 | ----------------- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 725 | |
Barry Warsaw | 52acb49 | 2001-12-21 20:04:22 +0000 | [diff] [blame] | 726 | - binascii.b2a_base64(): fix a potential buffer overrun when encoding |
| 727 | very short strings. |
| 728 | |
| 729 | - cPickle: the obscure "fast" mode was suspected of causing stack |
| 730 | overflows on the Mac. Hopefully fixed this by setting the recursion |
| 731 | limit much smaller. If the limit is too low (it only affects |
| 732 | performance), you can change it by defining PY_CPICKLE_FAST_LIMIT |
| 733 | when compiling cPickle.c (or in pyconfig.h). |
| 734 | |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 735 | Library |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 736 | ------- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 737 | |
Barry Warsaw | 52acb49 | 2001-12-21 20:04:22 +0000 | [diff] [blame] | 738 | - dumbdbm.py: fixed a dumb old bug (the file didn't get synched at |
| 739 | close or delete time). |
| 740 | |
| 741 | - rfc822.py: fixed a bug where the address '<>' was converted to None |
| 742 | instead of an empty string (also fixes the email.Utils module). |
| 743 | |
| 744 | - xmlrpclib.py: version 1.0.0; uses precision for doubles. |
| 745 | |
| 746 | - test suite: the pickle and cPickle tests were not executing any code |
Andrew M. Kuchling | 21b23b0 | 2002-05-07 20:58:03 +0000 | [diff] [blame] | 747 | when run from the standard regression test. |
Barry Warsaw | 52acb49 | 2001-12-21 20:04:22 +0000 | [diff] [blame] | 748 | |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 749 | Tools/Demos |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 750 | ----------- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 751 | |
| 752 | Build |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 753 | ----- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 754 | |
| 755 | C API |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 756 | ----- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 757 | |
| 758 | New platforms |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 759 | ------------- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 760 | |
| 761 | Tests |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 762 | ----- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 763 | |
| 764 | Windows |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 765 | ------- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 766 | |
Barry Warsaw | 52acb49 | 2001-12-21 20:04:22 +0000 | [diff] [blame] | 767 | - distutils package: fixed broken Windows installers (bdist_wininst). |
| 768 | |
| 769 | - tempfile.py: prevent mysterious warnings when TemporaryFileWrapper |
| 770 | instances are deleted at process exit time. |
| 771 | |
| 772 | - socket.py: prevent mysterious warnings when socket instances are |
| 773 | deleted at process exit time. |
| 774 | |
| 775 | - posixmodule.c: fix a Windows crash with stat() of a filename ending |
| 776 | in backslash. |
| 777 | |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 778 | Mac |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 779 | ---- |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 780 | |
Barry Warsaw | 52acb49 | 2001-12-21 20:04:22 +0000 | [diff] [blame] | 781 | - The Carbon toolbox modules have been upgraded to Universal Headers |
| 782 | 3.4, and experimental CoreGraphics and CarbonEvents modules have |
| 783 | been added. All only for framework-enabled MacOSX. |
| 784 | |
Tim Peters | 909c091 | 2001-12-14 23:16:18 +0000 | [diff] [blame] | 785 | |
Tim Peters | 6272dfd | 2001-12-13 19:34:00 +0000 | [diff] [blame] | 786 | What's New in Python 2.2c1? |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 787 | =========================== |
| 788 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 789 | *Release date: 14-Dec-2001* |
| 790 | |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 791 | Type/class unification and new-style classes |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 792 | -------------------------------------------- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 793 | |
Barry Warsaw | 0655745 | 2001-12-14 20:47:12 +0000 | [diff] [blame] | 794 | - Guido's tutorial introduction to the new type/class features has |
| 795 | been extensively updated. See |
| 796 | |
| 797 | http://www.python.org/2.2/descrintro.html |
| 798 | |
| 799 | That remains the primary documentation in this area. |
| 800 | |
Guido van Rossum | 33bab01 | 2001-12-05 22:45:48 +0000 | [diff] [blame] | 801 | - Fixed a leak: instance variables declared with __slots__ were never |
| 802 | deleted! |
| 803 | |
Guido van Rossum | 4b80085 | 2001-12-03 00:54:52 +0000 | [diff] [blame] | 804 | - The "delete attribute" method of descriptor objects is called |
| 805 | __delete__, not __del__. In previous releases, it was mistakenly |
| 806 | called __del__, which created an unfortunate overloading condition |
| 807 | with finalizers. (The "get attribute" and "set attribute" methods |
| 808 | are still called __get__ and __set__, respectively.) |
| 809 | |
Guido van Rossum | cdbbd0a | 2001-12-03 15:46:59 +0000 | [diff] [blame] | 810 | - Some subtle issues with the super built-in were fixed: |
| 811 | |
| 812 | (a) When super itself is subclassed, its __get__ method would still |
| 813 | return an instance of the base class (i.e., of super). |
| 814 | |
| 815 | (b) super(C, C()).__class__ would return C rather than super. This |
| 816 | is confusing. To fix this, I decided to change the semantics of |
| 817 | super so that it only applies to code attributes, not to data |
| 818 | attributes. After all, overriding data attributes is not |
| 819 | supported anyway. |
| 820 | |
| 821 | (c) The __get__ method didn't check whether the argument was an |
| 822 | instance of the type used in creation of the super instance. |
| 823 | |
Guido van Rossum | cd7b7d6 | 2001-12-04 21:02:07 +0000 | [diff] [blame] | 824 | - Previously, hash() of an instance of a subclass of a mutable type |
| 825 | (list or dictionary) would return some value, rather than raising |
| 826 | TypeError. This has been fixed. Also, directly calling |
| 827 | dict.__hash__ and list.__hash__ now raises the same TypeError |
| 828 | (previously, these were the same as object.__hash__). |
Guido van Rossum | cdbbd0a | 2001-12-03 15:46:59 +0000 | [diff] [blame] | 829 | |
Guido van Rossum | d331cb5 | 2001-12-05 19:46:42 +0000 | [diff] [blame] | 830 | - New-style objects now support deleting their __dict__. This is for |
| 831 | all intents and purposes equivalent to assigning a brand new empty |
| 832 | dictionary, but saves space if the object is not used further. |
| 833 | |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 834 | Core and builtins |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 835 | ----------------- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 836 | |
Tim Peters | 3caca23 | 2001-12-06 06:23:26 +0000 | [diff] [blame] | 837 | - -Qnew now works as documented in PEP 238: when -Qnew is passed on |
| 838 | the command line, all occurrences of "/" use true division instead |
| 839 | of classic division. See the PEP for details. Note that "all" |
| 840 | means all instances in library and 3rd-party modules, as well as in |
| 841 | your own code. As the PEP says, -Qnew is intended for use only in |
| 842 | educational environments with control over the libraries in use. |
Tim Peters | bf72fe2 | 2001-12-11 21:43:14 +0000 | [diff] [blame] | 843 | Note that test_coercion.py in the standard Python test suite fails |
| 844 | under -Qnew; this is expected, and won't be repaired until true |
| 845 | division becomes the default (in the meantime, test_coercion is |
| 846 | testing the current rules). |
Tim Peters | 3caca23 | 2001-12-06 06:23:26 +0000 | [diff] [blame] | 847 | |
Fred Drake | 6c01e25 | 2001-12-14 17:08:12 +0000 | [diff] [blame] | 848 | - complex() now only allows the first argument to be a string |
| 849 | argument, and raises TypeError if either the second arg is a string |
| 850 | or if the second arg is specified when the first is a string. |
| 851 | |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 852 | Extension modules |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 853 | ----------------- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 854 | |
Martin v. Löwis | 560da62 | 2001-11-24 09:24:51 +0000 | [diff] [blame] | 855 | - gc.get_referents was renamed to gc.get_referrers. |
| 856 | |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 857 | Library |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 858 | ------- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 859 | |
Tim Peters | 25059d3 | 2001-12-07 20:35:43 +0000 | [diff] [blame] | 860 | - Functions in the os.spawn() family now release the global interpreter |
| 861 | lock around calling the platform spawn. They should always have done |
| 862 | this, but did not before 2.2c1. Multithreaded programs calling |
| 863 | an os.spawn function with P_WAIT will no longer block all Python threads |
| 864 | until the spawned program completes. It's possible that some programs |
| 865 | relies on blocking, although more likely by accident than by design. |
| 866 | |
Martin v. Löwis | 3a89b2b | 2001-11-25 14:35:58 +0000 | [diff] [blame] | 867 | - webbrowser defaults to netscape.exe on OS/2 now. |
| 868 | |
Martin v. Löwis | 652e191 | 2001-11-25 14:50:56 +0000 | [diff] [blame] | 869 | - Tix.ResizeHandle exposes detach_widget, hide, and show. |
| 870 | |
Martin v. Löwis | 714d2e2 | 2001-12-02 12:26:03 +0000 | [diff] [blame] | 871 | - The charset alias windows_1252 has been added. |
| 872 | |
Barry Warsaw | 0655745 | 2001-12-14 20:47:12 +0000 | [diff] [blame] | 873 | - types.StringTypes is a tuple containing the defined string types; |
| 874 | usually this will be (str, unicode), but if Python was compiled |
| 875 | without Unicode support it will be just (str,). |
| 876 | |
| 877 | - The pulldom and minidom modules were synchronized to PyXML. |
| 878 | |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 879 | Tools/Demos |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 880 | ----------- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 881 | |
Barry Warsaw | 0655745 | 2001-12-14 20:47:12 +0000 | [diff] [blame] | 882 | - A new script called Tools/scripts/google.py was added, which fires |
| 883 | off a search on Google. |
| 884 | |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 885 | Build |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 886 | ----- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 887 | |
Barry Warsaw | 0655745 | 2001-12-14 20:47:12 +0000 | [diff] [blame] | 888 | - Note that release builds of Python should arrange to define the |
| 889 | preprocessor symbol NDEBUG on the command line (or equivalent). |
| 890 | In the 2.2 pre-release series we tried to define this by magic in |
| 891 | Python.h instead, but it proved to cause problems for extension |
| 892 | authors. The Unix, Windows and Mac builds now all define NDEBUG in |
| 893 | release builds via cmdline (or equivalent) instead. Ports to |
| 894 | other platforms should do likewise. |
| 895 | |
Jack Jansen | 1999ef4 | 2001-12-06 21:47:20 +0000 | [diff] [blame] | 896 | - It is no longer necessary to use --with-suffix when building on a |
| 897 | case-insensitive file system (such as Mac OS X HFS+). In the build |
| 898 | directory an extension is used, but not in the installed python. |
| 899 | |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 900 | C API |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 901 | ----- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 902 | |
Tim Peters | f582b82 | 2001-12-11 18:51:08 +0000 | [diff] [blame] | 903 | - New function PyDict_MergeFromSeq2() exposes the builtin dict |
| 904 | constructor's logic for updating a dictionary from an iterable object |
| 905 | producing key-value pairs. |
| 906 | |
Tim Peters | cffed4b | 2001-11-29 03:26:37 +0000 | [diff] [blame] | 907 | - PyArg_ParseTupleAndKeywords() requires that the number of entries in |
Tim Peters | f582b82 | 2001-12-11 18:51:08 +0000 | [diff] [blame] | 908 | the keyword list equal the number of argument specifiers. This |
Tim Peters | cffed4b | 2001-11-29 03:26:37 +0000 | [diff] [blame] | 909 | wasn't checked correctly, and PyArg_ParseTupleAndKeywords could even |
| 910 | dump core in some bad cases. This has been repaired. As a result, |
| 911 | PyArg_ParseTupleAndKeywords may raise RuntimeError in bad cases that |
| 912 | previously went unchallenged. |
| 913 | |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 914 | New platforms |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 915 | ------------- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 916 | |
| 917 | Tests |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 918 | ----- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 919 | |
| 920 | Windows |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 921 | ------- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 922 | |
| 923 | Mac |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 924 | ---- |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 925 | |
Jack Jansen | ffa260f | 2001-12-06 21:14:00 +0000 | [diff] [blame] | 926 | - In unix-Python on Mac OS X (and darwin) sys.platform is now "darwin", |
| 927 | without any trailing digits. |
Tim Peters | b3c9af1 | 2001-11-17 00:21:57 +0000 | [diff] [blame] | 928 | |
Barry Warsaw | 0655745 | 2001-12-14 20:47:12 +0000 | [diff] [blame] | 929 | - Changed logic for finding python home in Mac OS X framework Pythons. |
| 930 | Now sys.executable points to the executable again, in stead of to |
| 931 | the shared library. The latter is used only for locating the python |
| 932 | home. |
| 933 | |
| 934 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 935 | What's New in Python 2.2b2? |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 936 | =========================== |
| 937 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 938 | *Release date: 16-Nov-2001* |
| 939 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 940 | Type/class unification and new-style classes |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 941 | -------------------------------------------- |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 942 | |
Guido van Rossum | 3d27df0 | 2001-11-15 20:27:54 +0000 | [diff] [blame] | 943 | - Multiple inheritance mixing new-style and classic classes in the |
| 944 | list of base classes is now allowed, so this works now: |
Tim Peters | 76f7fe3 | 2001-11-15 20:02:21 +0000 | [diff] [blame] | 945 | |
| 946 | class Classic: pass |
Guido van Rossum | 3d27df0 | 2001-11-15 20:27:54 +0000 | [diff] [blame] | 947 | class Mixed(Classic, object): pass |
Tim Peters | 76f7fe3 | 2001-11-15 20:02:21 +0000 | [diff] [blame] | 948 | |
| 949 | The MRO (method resolution order) for each base class is respected |
| 950 | according to its kind, but the MRO for the derived class is computed |
Andrew M. Kuchling | 21b23b0 | 2002-05-07 20:58:03 +0000 | [diff] [blame] | 951 | using new-style MRO rules if any base class is a new-style class. |
Tim Peters | 76f7fe3 | 2001-11-15 20:02:21 +0000 | [diff] [blame] | 952 | This needs to be documented. |
| 953 | |
Tim Peters | a427a2b | 2001-10-29 22:25:45 +0000 | [diff] [blame] | 954 | - The new builtin dictionary() constructor, and dictionary type, have |
| 955 | been renamed to dict. This reflects a decade of common usage. |
| 956 | |
Guido van Rossum | 3fc08d2 | 2001-11-15 20:33:10 +0000 | [diff] [blame] | 957 | - dict() now accepts an iterable object producing 2-sequences. For |
| 958 | example, dict(d.items()) == d for any dictionary d. The argument, |
| 959 | and the elements of the argument, can be any iterable objects. |
| 960 | |
Guido van Rossum | 7ad2d1e | 2001-10-29 22:11:00 +0000 | [diff] [blame] | 961 | - New-style classes can now have a __del__ method, which is called |
| 962 | when the instance is deleted (just like for classic classes). |
| 963 | |
Guido van Rossum | 12d955c | 2001-10-26 14:56:06 +0000 | [diff] [blame] | 964 | - Assignment to object.__dict__ is now possible, for objects that are |
| 965 | instances of new-style classes that have a __dict__ (unless the base |
| 966 | class forbids it). |
| 967 | |
Guido van Rossum | c8e5645 | 2001-10-22 00:43:43 +0000 | [diff] [blame] | 968 | - Methods of built-in types now properly check for keyword arguments |
| 969 | (formerly these were silently ignored). The only built-in methods |
| 970 | that take keyword arguments are __call__, __init__ and __new__. |
| 971 | |
Guido van Rossum | 8644321 | 2001-10-27 22:28:54 +0000 | [diff] [blame] | 972 | - The socket function has been converted to a type; see below. |
| 973 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 974 | Core and builtins |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 975 | ----------------- |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 976 | |
Jeremy Hylton | 249aaed | 2001-11-09 21:06:24 +0000 | [diff] [blame] | 977 | - Assignment to __debug__ raises SyntaxError at compile-time. This |
| 978 | was promised when 2.1c1 was released as "What's New in Python 2.1c1" |
Barry Warsaw | 012852b | 2001-11-16 16:17:27 +0000 | [diff] [blame] | 979 | (see below) says. |
Jeremy Hylton | 249aaed | 2001-11-09 21:06:24 +0000 | [diff] [blame] | 980 | |
Guido van Rossum | 12d955c | 2001-10-26 14:56:06 +0000 | [diff] [blame] | 981 | - Clarified the error messages for unsupported operands to an operator |
| 982 | (like 1 + ''). |
| 983 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 984 | Extension modules |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 985 | ----------------- |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 986 | |
Tim Peters | 5ebfd36 | 2001-11-13 23:11:19 +0000 | [diff] [blame] | 987 | - mmap has a new keyword argument, "access", allowing a uniform way for |
| 988 | both Windows and Unix users to create read-only, write-through and |
| 989 | copy-on-write memory mappings. This was previously possible only on |
| 990 | Unix. A new keyword argument was required to support this in a |
Andrew M. Kuchling | 21b23b0 | 2002-05-07 20:58:03 +0000 | [diff] [blame] | 991 | uniform way because the mmap() signatures had diverged across |
Tim Peters | 5ebfd36 | 2001-11-13 23:11:19 +0000 | [diff] [blame] | 992 | platforms. Thanks to Jay T Miller for repairing this! |
| 993 | |
Tim Peters | 169ded0 | 2001-11-03 19:57:21 +0000 | [diff] [blame] | 994 | - By default, the gc.garbage list now contains only those instances in |
| 995 | unreachable cycles that have __del__ methods; in 2.1 it contained all |
| 996 | instances in unreachable cycles. "Instances" here has been generalized |
| 997 | to include instances of both new-style and old-style classes. |
| 998 | |
Guido van Rossum | 12d955c | 2001-10-26 14:56:06 +0000 | [diff] [blame] | 999 | - The socket module defines a new method for socket objects, |
| 1000 | sendall(). This is like send() but may make multiple calls to |
Guido van Rossum | 8644321 | 2001-10-27 22:28:54 +0000 | [diff] [blame] | 1001 | send() until all data has been sent. Also, the socket function has |
| 1002 | been converted to a subclassable type, like list and tuple (etc.) |
| 1003 | before it; socket and SocketType are now the same thing. |
Guido van Rossum | 12d955c | 2001-10-26 14:56:06 +0000 | [diff] [blame] | 1004 | |
| 1005 | - Various bugfixes to the curses module. There is now a test suite |
| 1006 | for the curses module (you have to run it manually). |
Andrew M. Kuchling | 15e0353 | 2001-10-22 16:37:10 +0000 | [diff] [blame] | 1007 | |
Guido van Rossum | 692adf1 | 2001-10-30 03:03:03 +0000 | [diff] [blame] | 1008 | - binascii.b2a_base64 no longer places an arbitrary restriction of 57 |
| 1009 | bytes on its input. |
| 1010 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1011 | Library |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1012 | ------- |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1013 | |
Tim Peters | 5ebfd36 | 2001-11-13 23:11:19 +0000 | [diff] [blame] | 1014 | - tkFileDialog exposes a Directory class and askdirectory |
Martin v. Löwis | 25ee87c | 2001-11-07 22:38:08 +0000 | [diff] [blame] | 1015 | convenience function. |
| 1016 | |
Tim Peters | 7533587 | 2001-11-03 19:35:43 +0000 | [diff] [blame] | 1017 | - Symbolic group names in regular expressions must be unique. For |
| 1018 | example, the regexp r'(?P<abc>)(?P<abc>)' is not allowed, because a |
| 1019 | single name can't mean both "group 1" and "group 2" simultaneously. |
Barry Warsaw | 012852b | 2001-11-16 16:17:27 +0000 | [diff] [blame] | 1020 | Python 2.2 detects this error at regexp compilation time; |
| 1021 | previously, the error went undetected, and results were |
| 1022 | unpredictable. Also in sre, the pattern.split(), pattern.sub(), and |
| 1023 | pattern.subn() methods have been rewritten in C. Also, an |
| 1024 | experimental function/method finditer() has been added, which works |
| 1025 | like findall() but returns an iterator. |
Tim Peters | 7533587 | 2001-11-03 19:35:43 +0000 | [diff] [blame] | 1026 | |
Martin v. Löwis | b7b3260 | 2001-11-02 23:48:20 +0000 | [diff] [blame] | 1027 | - Tix exposes more commands through the classes DirSelectBox, |
| 1028 | DirSelectDialog, ListNoteBook, Meter, CheckList, and the |
| 1029 | methods tix_addbitmapdir, tix_cget, tix_configure, tix_filedialog, |
| 1030 | tix_getbitmap, tix_getimage, tix_option_get, and tix_resetoptions. |
| 1031 | |
Tim Peters | 98791af | 2001-10-23 01:59:54 +0000 | [diff] [blame] | 1032 | - Traceback objects are now scanned by cyclic garbage collection, so |
| 1033 | cycles created by casual use of sys.exc_info() no longer cause |
| 1034 | permanent memory leaks (provided garbage collection is enabled). |
| 1035 | |
Guido van Rossum | 4919214 | 2001-10-24 20:51:44 +0000 | [diff] [blame] | 1036 | - os.extsep -- a new variable needed by the RISCOS support. It is the |
| 1037 | separator used by extensions, and is '.' on all platforms except |
| 1038 | RISCOS, where it is '/'. There is no need to use this variable |
| 1039 | unless you have a masochistic desire to port your code to RISCOS. |
| 1040 | |
Barry Warsaw | 012852b | 2001-11-16 16:17:27 +0000 | [diff] [blame] | 1041 | - mimetypes.py has optional support for non-standard, but commonly |
| 1042 | found types. guess_type() and guess_extension() now accept an |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1043 | optional 'strict' flag, defaulting to true, which controls whether |
Barry Warsaw | 012852b | 2001-11-16 16:17:27 +0000 | [diff] [blame] | 1044 | recognize non-standard types or not. A few non-standard types we |
| 1045 | know about have been added. Also, when run as a script, there are |
| 1046 | new -l and -e options. |
| 1047 | |
| 1048 | - statcache is now deprecated. |
| 1049 | |
| 1050 | - email.Utils.formatdate() now produces the preferred RFC 2822 style |
| 1051 | dates with numeric timezones (it used to produce obsolete dates |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1052 | hard coded to "GMT" timezone). An optional 'localtime' flag is |
Barry Warsaw | 012852b | 2001-11-16 16:17:27 +0000 | [diff] [blame] | 1053 | added to produce dates in the local timezone, with daylight savings |
| 1054 | time properly taken into account. |
| 1055 | |
| 1056 | - In pickle and cPickle, instead of masking errors in load() by |
| 1057 | transforming them into SystemError, we let the original exception |
| 1058 | propagate out. Also, implement support for __safe_for_unpickling__ |
| 1059 | in pickle, as it already was supported in cPickle. |
| 1060 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1061 | Tools/Demos |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1062 | ----------- |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1063 | |
| 1064 | Build |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1065 | ----- |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1066 | |
Barry Warsaw | 012852b | 2001-11-16 16:17:27 +0000 | [diff] [blame] | 1067 | - The dbm module is built using libdb1 if available. The bsddb module |
| 1068 | is built with libdb3 if available. |
| 1069 | |
| 1070 | - Misc/Makefile.pre.in has been removed by BDFL pronouncement. |
| 1071 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1072 | C API |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1073 | ----- |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1074 | |
Tim Peters | 1fc240e | 2001-10-26 05:06:50 +0000 | [diff] [blame] | 1075 | - New function PySequence_Fast_GET_SIZE() returns the size of a non- |
| 1076 | NULL result from PySequence_Fast(), more quickly than calling |
| 1077 | PySequence_Size(). |
| 1078 | |
Barry Warsaw | 012852b | 2001-11-16 16:17:27 +0000 | [diff] [blame] | 1079 | - New argument unpacking function PyArg_UnpackTuple() added. |
| 1080 | |
| 1081 | - New functions PyObject_CallFunctionObjArgs() and |
| 1082 | PyObject_CallMethodObjArgs() have been added to make it more |
| 1083 | convenient and efficient to call functions and methods from C. |
| 1084 | |
| 1085 | - PyArg_ParseTupleAndKeywords() no longer masks errors, so it's |
| 1086 | possible that this will propagate errors it didn't before. |
| 1087 | |
| 1088 | - New function PyObject_CheckReadBuffer(), which returns true if its |
| 1089 | argument supports the single-segment readable buffer interface. |
| 1090 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1091 | New platforms |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1092 | ------------- |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1093 | |
Guido van Rossum | 8644321 | 2001-10-27 22:28:54 +0000 | [diff] [blame] | 1094 | - We've finally confirmed that this release builds on HP-UX 11.00, |
| 1095 | *with* threads, and passes the test suite. |
| 1096 | |
Tim Peters | 7a33d8b | 2001-11-05 02:51:07 +0000 | [diff] [blame] | 1097 | - Thanks to a series of patches from Michael Muller, Python may build |
| 1098 | again under OS/2 Visual Age C++. |
| 1099 | |
Guido van Rossum | 4919214 | 2001-10-24 20:51:44 +0000 | [diff] [blame] | 1100 | - Updated RISCOS port by Dietmar Schwertberger. |
| 1101 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1102 | Tests |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1103 | ----- |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1104 | |
Andrew M. Kuchling | 15e0353 | 2001-10-22 16:37:10 +0000 | [diff] [blame] | 1105 | - Added a test script for the curses module. It isn't run automatically; |
| 1106 | regrtest.py must be run with '-u curses' to enable it. |
| 1107 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1108 | Windows |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1109 | ------- |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1110 | |
Barry Warsaw | 012852b | 2001-11-16 16:17:27 +0000 | [diff] [blame] | 1111 | Mac |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1112 | ---- |
Barry Warsaw | 012852b | 2001-11-16 16:17:27 +0000 | [diff] [blame] | 1113 | |
| 1114 | - PythonScript has been moved to unsupported and is slated to be |
| 1115 | removed completely in the next release. |
| 1116 | |
| 1117 | - It should now be possible to build applets that work on both OS9 and |
| 1118 | OSX. |
| 1119 | |
| 1120 | - The core is now linked with CoreServices not Carbon; as a side |
| 1121 | result, default 8bit encoding on OSX is now ASCII. |
| 1122 | |
| 1123 | - Python should now build on OSX 10.1.1 |
| 1124 | |
Tim Peters | 84362bc | 2001-10-19 17:55:30 +0000 | [diff] [blame] | 1125 | |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1126 | What's New in Python 2.2b1? |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1127 | =========================== |
| 1128 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1129 | *Release date: 19-Oct-2001* |
| 1130 | |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1131 | Type/class unification and new-style classes |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1132 | -------------------------------------------- |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1133 | |
Guido van Rossum | 687ae00 | 2001-10-15 22:03:32 +0000 | [diff] [blame] | 1134 | - New-style classes are now always dynamic (except for built-in and |
Guido van Rossum | 3eea25c | 2001-10-16 00:46:57 +0000 | [diff] [blame] | 1135 | extension types). There is no longer a performance penalty, and I |
Guido van Rossum | 687ae00 | 2001-10-15 22:03:32 +0000 | [diff] [blame] | 1136 | no longer see another reason to keep this baggage around. One relic |
Guido van Rossum | 3eea25c | 2001-10-16 00:46:57 +0000 | [diff] [blame] | 1137 | remains: the __dict__ of a new-style class is a read-only proxy; you |
| 1138 | must set the class's attribute to modify it. As a consequence, the |
Guido van Rossum | 687ae00 | 2001-10-15 22:03:32 +0000 | [diff] [blame] | 1139 | __defined__ attribute of new-style types no longer exists, for lack |
| 1140 | of need: there is once again only one __dict__ (although in the |
Guido van Rossum | 3eea25c | 2001-10-16 00:46:57 +0000 | [diff] [blame] | 1141 | future a __cache__ may be resurrected with a similar function, if I |
| 1142 | can prove that it actually speeds things up). |
Guido van Rossum | 50fda3b | 2001-10-04 19:46:06 +0000 | [diff] [blame] | 1143 | |
Tim Peters | 1c9ca87 | 2001-10-04 06:43:12 +0000 | [diff] [blame] | 1144 | - C.__doc__ now works as expected for new-style classes (in 2.2a4 it |
| 1145 | always returned None, even when there was a class docstring). |
| 1146 | |
| 1147 | - doctest now finds and runs docstrings attached to new-style classes, |
| 1148 | class methods, static methods, and properties. |
| 1149 | |
Guido van Rossum | 9074ef6 | 2001-10-16 21:34:49 +0000 | [diff] [blame] | 1150 | Core and builtins |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1151 | ----------------- |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1152 | |
Guido van Rossum | a8bcf80 | 2001-10-15 15:53:58 +0000 | [diff] [blame] | 1153 | - A very subtle syntactical pitfall in list comprehensions was fixed. |
| 1154 | For example: [a+b for a in 'abc', for b in 'def']. The comma in |
| 1155 | this example is a mistake. Previously, this would silently let 'a' |
| 1156 | iterate over the singleton tuple ('abc',), yielding ['abcd', 'abce', |
| 1157 | 'abcf'] rather than the intended ['ad', 'ae', 'af', 'bd', 'be', |
| 1158 | 'bf', 'cd', 'ce', 'cf']. Now, this is flagged as a syntax error. |
| 1159 | Note that [a for a in <singleton>] is a convoluted way to say |
| 1160 | [<singleton>] anyway, so it's not like any expressiveness is lost. |
| 1161 | |
Guido van Rossum | 9074ef6 | 2001-10-16 21:34:49 +0000 | [diff] [blame] | 1162 | - getattr(obj, name, default) now only catches AttributeError, as |
| 1163 | documented, rather than returning the default value for all |
| 1164 | exceptions (which could mask bugs in a __getattr__ hook, for |
| 1165 | example). |
| 1166 | |
Fred Drake | cb7a6b5 | 2001-10-22 18:41:51 +0000 | [diff] [blame] | 1167 | - Weak reference objects are now part of the core and offer a C API. |
Fred Drake | 934d2a4 | 2001-10-18 18:18:06 +0000 | [diff] [blame] | 1168 | A bug which could allow a core dump when binary operations involved |
Andrew M. Kuchling | 51c1816 | 2001-10-22 01:47:26 +0000 | [diff] [blame] | 1169 | proxy reference has been fixed. weakref.ReferenceError is now a |
Barry Warsaw | f6fb171 | 2001-10-19 05:35:40 +0000 | [diff] [blame] | 1170 | built-in exception. |
Fred Drake | 934d2a4 | 2001-10-18 18:18:06 +0000 | [diff] [blame] | 1171 | |
Guido van Rossum | 1c48609 | 2001-10-19 02:05:35 +0000 | [diff] [blame] | 1172 | - unicode(obj) now behaves more like str(obj), accepting arbitrary |
| 1173 | objects, and calling a __unicode__ method if it exists. |
| 1174 | unicode(obj, encoding) and unicode(obj, encoding, errors) still |
Marc-André Lemburg | b5507ec | 2001-10-19 12:02:29 +0000 | [diff] [blame] | 1175 | require an 8-bit string or character buffer argument. |
Guido van Rossum | 1c48609 | 2001-10-19 02:05:35 +0000 | [diff] [blame] | 1176 | |
Barry Warsaw | f6fb171 | 2001-10-19 05:35:40 +0000 | [diff] [blame] | 1177 | - isinstance() now allows any object as the first argument and a |
| 1178 | class, a type or something with a __bases__ tuple attribute for the |
| 1179 | second argument. The second argument may also be a tuple of a |
| 1180 | class, type, or something with __bases__, in which case isinstance() |
| 1181 | will return true if the first argument is an instance of any of the |
| 1182 | things contained in the second argument tuple. E.g. |
| 1183 | |
| 1184 | isinstance(x, (A, B)) |
| 1185 | |
| 1186 | returns true if x is an instance of A or B. |
| 1187 | |
Guido van Rossum | 9074ef6 | 2001-10-16 21:34:49 +0000 | [diff] [blame] | 1188 | Extension modules |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1189 | ----------------- |
Guido van Rossum | 9074ef6 | 2001-10-16 21:34:49 +0000 | [diff] [blame] | 1190 | |
| 1191 | - thread.start_new_thread() now returns the thread ID (previously None). |
| 1192 | |
Martin v. Löwis | 16dc7f4 | 2001-09-30 20:32:11 +0000 | [diff] [blame] | 1193 | - binascii has now two quopri support functions, a2b_qp and b2a_qp. |
| 1194 | |
Guido van Rossum | c524d95 | 2001-10-19 01:31:59 +0000 | [diff] [blame] | 1195 | - readline now supports setting the startup_hook and the |
| 1196 | pre_event_hook, and adds the add_history() function. |
Martin v. Löwis | 0daad59 | 2001-09-30 21:09:59 +0000 | [diff] [blame] | 1197 | |
Guido van Rossum | c524d95 | 2001-10-19 01:31:59 +0000 | [diff] [blame] | 1198 | - os and posix supports chroot(), setgroups() and unsetenv() where |
| 1199 | available. The stat(), fstat(), statvfs() and fstatvfs() functions |
| 1200 | now return "pseudo-sequences" -- the various fields can now be |
| 1201 | accessed as attributes (e.g. os.stat("/").st_mtime) but for |
| 1202 | backwards compatibility they also behave as a fixed-length sequence. |
| 1203 | Some platform-specific fields (e.g. st_rdev) are only accessible as |
Guido van Rossum | 98bf58f | 2001-10-18 20:34:25 +0000 | [diff] [blame] | 1204 | attributes. |
| 1205 | |
| 1206 | - time: localtime(), gmtime() and strptime() now return a |
| 1207 | pseudo-sequence similar to the os.stat() return value, with |
| 1208 | attributes like tm_year etc. |
Martin v. Löwis | 16628c9 | 2001-10-04 22:46:41 +0000 | [diff] [blame] | 1209 | |
Guido van Rossum | 9074ef6 | 2001-10-16 21:34:49 +0000 | [diff] [blame] | 1210 | - Decompression objects in the zlib module now accept an optional |
| 1211 | second parameter to decompress() that specifies the maximum amount |
| 1212 | of memory to use for the uncompressed data. |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1213 | |
Jeremy Hylton | cb44143 | 2001-10-19 03:40:19 +0000 | [diff] [blame] | 1214 | - optional SSL support in the socket module now exports OpenSSL |
| 1215 | functions RAND_add(), RAND_egd(), and RAND_status(). These calls |
| 1216 | are useful on platforms like Solaris where OpenSSL does not |
Barry Warsaw | f6fb171 | 2001-10-19 05:35:40 +0000 | [diff] [blame] | 1217 | automatically seed its PRNG. Also, the keyfile and certfile |
| 1218 | arguments to socket.ssl() are now optional. |
| 1219 | |
| 1220 | - posixmodule (and by extension, the os module on POSIX platforms) now |
| 1221 | exports O_LARGEFILE, O_DIRECT, O_DIRECTORY, and O_NOFOLLOW. |
Jeremy Hylton | cb44143 | 2001-10-19 03:40:19 +0000 | [diff] [blame] | 1222 | |
Guido van Rossum | 9074ef6 | 2001-10-16 21:34:49 +0000 | [diff] [blame] | 1223 | Library |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1224 | ------- |
Guido van Rossum | 3c28863 | 2001-10-16 21:13:49 +0000 | [diff] [blame] | 1225 | |
Tim Peters | 7402f79 | 2001-10-02 03:53:41 +0000 | [diff] [blame] | 1226 | - doctest now excludes functions and classes not defined by the module |
| 1227 | being tested, thanks to Tim Hochberg. |
| 1228 | |
Fred Drake | cb7a6b5 | 2001-10-22 18:41:51 +0000 | [diff] [blame] | 1229 | - HotShot, a new profiler implemented using a C-based callback, has |
| 1230 | been added. This substantially reduces the overhead of profiling, |
| 1231 | but it is still quite preliminary. Support modules and |
| 1232 | documentation will be added in upcoming releases (before 2.2 final). |
| 1233 | |
Guido van Rossum | c4b09b4 | 2001-10-04 10:19:00 +0000 | [diff] [blame] | 1234 | - profile now produces correct output in situations where an exception |
| 1235 | raised in Python is cleared by C code (e.g. hasattr()). This used |
| 1236 | to cause wrong output, including spurious claims of recursive |
| 1237 | functions and attribution of time spent to the wrong function. |
| 1238 | |
Tim Peters | 0a1fc4e | 2001-10-07 03:12:08 +0000 | [diff] [blame] | 1239 | The code and documentation for the derived OldProfile and HotProfile |
| 1240 | profiling classes was removed. The code hasn't worked for years (if |
| 1241 | you tried to use them, they raised exceptions). OldProfile |
| 1242 | intended to reproduce the behavior of the profiler Python used more |
| 1243 | than 7 years ago, and isn't interesting anymore. HotProfile intended |
| 1244 | to provide a faster profiler (but producing less information), and |
| 1245 | that's a worthy goal we intend to meet via a different approach (but |
| 1246 | without losing information). |
| 1247 | |
Tim Peters | cce092d | 2001-10-09 05:31:56 +0000 | [diff] [blame] | 1248 | - Profile.calibrate() has a new implementation that should deliver |
Tim Peters | 659a603 | 2001-10-09 20:51:19 +0000 | [diff] [blame] | 1249 | a much better system-specific calibration constant. The constant can |
| 1250 | now be specified in an instance constructor, or as a Profile class or |
| 1251 | instance variable, instead of by editing profile.py's source code. |
| 1252 | Calibration must still be done manually (see the docs for the profile |
| 1253 | module). |
| 1254 | |
| 1255 | Note that Profile.calibrate() must be overriden by subclasses. |
| 1256 | Improving the accuracy required exploiting detailed knowledge of |
| 1257 | profiler internals; the earlier method abstracted away the details |
| 1258 | and measured a simplified model instead, but consequently computed |
| 1259 | a constant too small by a factor of 2 on some modern machines. |
Tim Peters | cce092d | 2001-10-09 05:31:56 +0000 | [diff] [blame] | 1260 | |
Martin v. Löwis | 16dc7f4 | 2001-09-30 20:32:11 +0000 | [diff] [blame] | 1261 | - quopri's encode and decode methods take an optional header parameter, |
Jeremy Hylton | 6f543b6 | 2001-10-16 20:42:52 +0000 | [diff] [blame] | 1262 | which indicates whether output is intended for the header 'Q' |
| 1263 | encoding. |
| 1264 | |
Guido van Rossum | a5343cc | 2001-10-18 18:02:07 +0000 | [diff] [blame] | 1265 | - The SocketServer.ThreadingMixIn class now closes the request after |
| 1266 | finish_request() returns. (Not when it errors out though.) |
| 1267 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1268 | - The nntplib module's NNTP.body() method has grown a 'file' argument |
Barry Warsaw | f6fb171 | 2001-10-19 05:35:40 +0000 | [diff] [blame] | 1269 | to allow saving the message body to a file. |
| 1270 | |
| 1271 | - The email package has added a class email.Parser.HeaderParser which |
| 1272 | only parses headers and does not recurse into the message's body. |
| 1273 | Also, the module/class MIMEAudio has been added for representing |
| 1274 | audio data (contributed by Anthony Baxter). |
| 1275 | |
| 1276 | - ftplib should be able to handle files > 2GB. |
| 1277 | |
| 1278 | - ConfigParser.getboolean() now also interprets TRUE, FALSE, YES, NO, |
| 1279 | ON, and OFF. |
| 1280 | |
Fred Drake | cb7a6b5 | 2001-10-22 18:41:51 +0000 | [diff] [blame] | 1281 | - xml.dom.minidom NodeList objects now support the length attribute |
| 1282 | and item() method as required by the DOM specifications. |
| 1283 | |
Guido van Rossum | ed554f6 | 2001-10-02 23:15:37 +0000 | [diff] [blame] | 1284 | Tools/Demos |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1285 | ----------- |
Guido van Rossum | ed554f6 | 2001-10-02 23:15:37 +0000 | [diff] [blame] | 1286 | |
| 1287 | - Demo/dns was removed. It no longer serves any purpose; a package |
| 1288 | derived from it is now maintained by Anthony Baxter, see |
| 1289 | http://PyDNS.SourceForge.net. |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1290 | |
Barry Warsaw | f6fb171 | 2001-10-19 05:35:40 +0000 | [diff] [blame] | 1291 | - The freeze tool has been made more robust, and two new options have |
| 1292 | been added: -X and -E. |
| 1293 | |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1294 | Build |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1295 | ----- |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1296 | |
Barry Warsaw | f6fb171 | 2001-10-19 05:35:40 +0000 | [diff] [blame] | 1297 | - configure will use CXX in LINKCC if CXX is used to build main() and |
| 1298 | the system requires to link a C++ main using the C++ compiler. |
| 1299 | |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1300 | C API |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1301 | ----- |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1302 | |
Guido van Rossum | 6c4bce3 | 2001-10-18 19:20:25 +0000 | [diff] [blame] | 1303 | - The documentation for the tp_compare slot is updated to require that |
| 1304 | the return value must be -1, 0, 1; an arbitrary number <0 or >0 is |
| 1305 | not correct. This is not yet enforced but will be enforced in |
| 1306 | Python 2.3; even later, we may use -2 to indicate errors and +2 for |
| 1307 | "NotImplemented". Right now, -1 should be used for an error return. |
| 1308 | |
Tim Peters | d38b1c7 | 2001-09-30 05:09:37 +0000 | [diff] [blame] | 1309 | - PyLong_AsLongLong() now accepts int (as well as long) arguments. |
| 1310 | Consequently, PyArg_ParseTuple's 'L' code also accepts int (as well |
| 1311 | as long) arguments. |
| 1312 | |
Guido van Rossum | 3c28863 | 2001-10-16 21:13:49 +0000 | [diff] [blame] | 1313 | - PyThread_start_new_thread() now returns a long int giving the thread |
| 1314 | ID, if one can be calculated; it returns -1 for error, 0 if no |
| 1315 | thread ID is calculated (this is an incompatible change, but only |
| 1316 | the thread module used this API). This code has only really been |
| 1317 | tested on Linux and Windows; other platforms please beware (and |
| 1318 | report any bugs or strange behavior). |
| 1319 | |
Guido van Rossum | 1c48609 | 2001-10-19 02:05:35 +0000 | [diff] [blame] | 1320 | - PyUnicode_FromEncodedObject() no longer accepts Unicode objects as |
| 1321 | input. |
| 1322 | |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1323 | New platforms |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1324 | ------------- |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1325 | |
| 1326 | Tests |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1327 | ----- |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1328 | |
| 1329 | Windows |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1330 | ------- |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1331 | |
Tim Peters | 04cf1d3 | 2001-10-09 22:39:40 +0000 | [diff] [blame] | 1332 | - Installer: If you install IDLE, and don't disable file-extension |
| 1333 | registration, a new "Edit with IDLE" context (right-click) menu entry |
| 1334 | is created for .py and .pyw files. |
| 1335 | |
Tim Peters | 1ce3cf7 | 2001-10-01 17:58:40 +0000 | [diff] [blame] | 1336 | - The signal module now supports SIGBREAK on Windows, thanks to Steven |
| 1337 | Scott. Note that SIGBREAK is unique to Windows. The default SIGBREAK |
| 1338 | action remains to call Win32 ExitProcess(). This can be changed via |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1339 | signal.signal(). For example: :: |
Tim Peters | 1ce3cf7 | 2001-10-01 17:58:40 +0000 | [diff] [blame] | 1340 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1341 | # Make Ctrl+Break raise KeyboardInterrupt, like Python's default Ctrl+C |
| 1342 | # (SIGINT) behavior. |
| 1343 | import signal |
| 1344 | signal.signal(signal.SIGBREAK, signal.default_int_handler) |
Tim Peters | 1ce3cf7 | 2001-10-01 17:58:40 +0000 | [diff] [blame] | 1345 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1346 | try: |
| 1347 | while 1: |
| 1348 | pass |
| 1349 | except KeyboardInterrupt: |
| 1350 | # We get here on Ctrl+C or Ctrl+Break now; if we had not changed |
| 1351 | # SIGBREAK, only on Ctrl+C (and Ctrl+Break would terminate the |
| 1352 | # program without the possibility for any Python-level cleanup). |
| 1353 | print "Clean exit" |
Tim Peters | 1ce3cf7 | 2001-10-01 17:58:40 +0000 | [diff] [blame] | 1354 | |
Tim Peters | abf925f | 2001-09-28 21:53:42 +0000 | [diff] [blame] | 1355 | |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1356 | What's New in Python 2.2a4? |
| 1357 | =========================== |
| 1358 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1359 | *Release date: 28-Sep-2001* |
| 1360 | |
Guido van Rossum | 808eea7 | 2001-09-25 04:15:41 +0000 | [diff] [blame] | 1361 | Type/class unification and new-style classes |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1362 | -------------------------------------------- |
Guido van Rossum | 808eea7 | 2001-09-25 04:15:41 +0000 | [diff] [blame] | 1363 | |
| 1364 | - pydoc and inspect are now aware of new-style classes; |
| 1365 | e.g. help(list) at the interactive prompt now shows proper |
| 1366 | documentation for all operations on list objects. |
| 1367 | |
| 1368 | - Applications using Jim Fulton's ExtensionClass module can now safely |
| 1369 | be used with Python 2.2. In particular, Zope 2.4.1 now works with |
| 1370 | Python 2.2 (as well as with Python 2.1.1). The Demo/metaclass |
| 1371 | examples also work again. It is hoped that Gtk and Boost also work |
| 1372 | with 2.2a4 and beyond. (If you can confirm this, please write |
| 1373 | webmaster@python.org; if there are still problems, please open a bug |
| 1374 | report on SourceForge.) |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1375 | |
Tim Peters | 66c1a52 | 2001-09-24 21:17:50 +0000 | [diff] [blame] | 1376 | - property() now takes 4 keyword arguments: fget, fset, fdel and doc. |
Andrew M. Kuchling | 21b23b0 | 2002-05-07 20:58:03 +0000 | [diff] [blame] | 1377 | These map to read-only attributes 'fget', 'fset', 'fdel', and '__doc__' |
Tim Peters | 66c1a52 | 2001-09-24 21:17:50 +0000 | [diff] [blame] | 1378 | in the constructed property object. fget, fset and fdel weren't |
| 1379 | discoverable from Python in 2.2a3. __doc__ is new, and allows to |
| 1380 | associate a docstring with a property. |
| 1381 | |
Guido van Rossum | 808eea7 | 2001-09-25 04:15:41 +0000 | [diff] [blame] | 1382 | - Comparison overloading is now more completely implemented. For |
| 1383 | example, a str subclass instance can properly be compared to a str |
| 1384 | instance, and it can properly overload comparison. Ditto for most |
| 1385 | other built-in object types. |
| 1386 | |
| 1387 | - The repr() of new-style classes has changed; instead of <type |
| 1388 | 'M.Foo'> a new-style class is now rendered as <class 'M.Foo'>, |
| 1389 | *except* for built-in types, which are still rendered as <type |
| 1390 | 'Foo'> (to avoid upsetting existing code that might parse or |
| 1391 | otherwise rely on repr() of certain type objects). |
| 1392 | |
| 1393 | - The repr() of new-style objects is now always <Foo object at XXX>; |
| 1394 | previously, it was sometimes <Foo instance at XXX>. |
| 1395 | |
| 1396 | - For new-style classes, what was previously called __getattr__ is now |
| 1397 | called __getattribute__. This method, if defined, is called for |
Andrew M. Kuchling | 51c1816 | 2001-10-22 01:47:26 +0000 | [diff] [blame] | 1398 | *every* attribute access. A new __getattr__ hook more similar to the |
Guido van Rossum | 808eea7 | 2001-09-25 04:15:41 +0000 | [diff] [blame] | 1399 | one in classic classes is defined which is called only if regular |
| 1400 | attribute access raises AttributeError; to catch *all* attribute |
| 1401 | access, you can use __getattribute__ (for new-style classes). If |
| 1402 | both are defined, __getattribute__ is called first, and if it raises |
| 1403 | AttributeError, __getattr__ is called. |
| 1404 | |
| 1405 | - The __class__ attribute of new-style objects can be assigned to. |
| 1406 | The new class must have the same C-level object layout as the old |
| 1407 | class. |
| 1408 | |
| 1409 | - The builtin file type can be subclassed now. In the usual pattern, |
| 1410 | "file" is the name of the builtin type, and file() is a new builtin |
| 1411 | constructor, with the same signature as the builtin open() function. |
| 1412 | file() is now the preferred way to open a file. |
| 1413 | |
| 1414 | - Previously, __new__ would only see sequential arguments passed to |
| 1415 | the type in a constructor call; __init__ would see both sequential |
| 1416 | and keyword arguments. This made no sense whatsoever any more, so |
| 1417 | now both __new__ and __init__ see all arguments. |
| 1418 | |
| 1419 | - Previously, hash() applied to an instance of a subclass of str or |
| 1420 | unicode always returned 0. This has been repaired. |
| 1421 | |
| 1422 | - Previously, an operation on an instance of a subclass of an |
| 1423 | immutable type (int, long, float, complex, tuple, str, unicode), |
| 1424 | where the subtype didn't override the operation (and so the |
| 1425 | operation was handled by the builtin type), could return that |
| 1426 | instance instead a value of the base type. For example, if s was of |
Andrew M. Kuchling | 21b23b0 | 2002-05-07 20:58:03 +0000 | [diff] [blame] | 1427 | a str subclass type, s[:] returned s as-is. Now it returns a str |
Guido van Rossum | 808eea7 | 2001-09-25 04:15:41 +0000 | [diff] [blame] | 1428 | with the same value as s. |
| 1429 | |
Barry Warsaw | 86fbaf8 | 2001-09-28 15:26:12 +0000 | [diff] [blame] | 1430 | - Provisional support for pickling new-style objects has been added. |
| 1431 | |
Guido van Rossum | 808eea7 | 2001-09-25 04:15:41 +0000 | [diff] [blame] | 1432 | Core |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1433 | ---- |
Guido van Rossum | 808eea7 | 2001-09-25 04:15:41 +0000 | [diff] [blame] | 1434 | |
Tim Peters | 2c9aa5e | 2001-09-23 04:06:05 +0000 | [diff] [blame] | 1435 | - file.writelines() now accepts any iterable object producing strings. |
| 1436 | |
Marc-André Lemburg | aefd766 | 2001-09-20 12:59:37 +0000 | [diff] [blame] | 1437 | - PyUnicode_FromEncodedObject() now works very much like |
| 1438 | PyObject_Str(obj) in that it tries to use __str__/tp_str |
| 1439 | on the object if the object is not a string or buffer. This |
| 1440 | makes unicode() behave like str() when applied to non-string/buffer |
| 1441 | objects. |
| 1442 | |
Andrew M. Kuchling | 51c1816 | 2001-10-22 01:47:26 +0000 | [diff] [blame] | 1443 | - PyFile_WriteObject now passes Unicode objects to the file's write |
| 1444 | method. As a result, all file-like objects which may be the target |
Martin v. Löwis | 2777c02 | 2001-09-19 13:47:32 +0000 | [diff] [blame] | 1445 | of a print statement must support Unicode objects, i.e. they must |
| 1446 | at least convert them into ASCII strings. |
| 1447 | |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1448 | - Thread scheduling on Solaris should be improved; it is no longer |
| 1449 | necessary to insert a small sleep at the start of a thread in order |
| 1450 | to let other runnable threads be scheduled. |
| 1451 | |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1452 | Library |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1453 | ------- |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1454 | |
Marc-André Lemburg | bf99017 | 2001-09-27 14:17:33 +0000 | [diff] [blame] | 1455 | - StringIO.StringIO instances and cStringIO.StringIO instances support |
| 1456 | read character buffer compatible objects for their .write() methods. |
| 1457 | These objects are converted to strings and then handled as such |
| 1458 | by the instances. |
| 1459 | |
Barry Warsaw | 2f60073 | 2001-09-24 04:28:10 +0000 | [diff] [blame] | 1460 | - The "email" package has been added. This is basically a port of the |
| 1461 | mimelib package <http://sf.net/projects/mimelib> with API changes |
| 1462 | and some implementations updated to use iterators and generators. |
| 1463 | |
Tim Peters | 8a9c284 | 2001-09-22 21:30:22 +0000 | [diff] [blame] | 1464 | - difflib.ndiff() and difflib.Differ.compare() are generators now. This |
| 1465 | restores the ability of Tools/scripts/ndiff.py to start producing output |
| 1466 | before the entire comparison is complete. |
| 1467 | |
Barry Warsaw | 58b072d | 2001-09-22 04:44:21 +0000 | [diff] [blame] | 1468 | - StringIO.StringIO instances and cStringIO.StringIO instances support |
| 1469 | iteration just like file objects (i.e. their .readline() method is |
| 1470 | called for each iteration until it returns an empty string). |
| 1471 | |
Marc-André Lemburg | 494f2ae | 2001-09-19 11:33:31 +0000 | [diff] [blame] | 1472 | - The codecs module has grown four new helper APIs to access |
| 1473 | builtin codecs: getencoder(), getdecoder(), getreader(), |
| 1474 | getwriter(). |
| 1475 | |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1476 | - SimpleXMLRPCServer: a new module (based upon SimpleHTMLServer) |
| 1477 | simplifies writing XML RPC servers. |
| 1478 | |
Barry Warsaw | 647d5e8 | 2001-09-28 17:01:02 +0000 | [diff] [blame] | 1479 | - os.path.realpath(): a new function that returns the absolute pathname |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1480 | after interpretation of symbolic links. On non-Unix systems, this |
| 1481 | is an alias for os.path.abspath(). |
| 1482 | |
Tim Peters | 16a77ad | 2001-09-08 04:00:12 +0000 | [diff] [blame] | 1483 | - operator.indexOf() (PySequence_Index() in the C API) now works with any |
| 1484 | iterable object. |
| 1485 | |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1486 | - smtplib now supports various authentication and security features of |
| 1487 | the SMTP protocol through the new login() and starttls() methods. |
Guido van Rossum | d8185ca | 2001-09-14 16:35:16 +0000 | [diff] [blame] | 1488 | |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1489 | - hmac: a new module implementing keyed hashing for message |
| 1490 | authentication. |
| 1491 | |
| 1492 | - mimetypes now recognizes more extensions and file types. At the |
| 1493 | same time, some mappings not sanctioned by IANA were removed. |
Guido van Rossum | d8185ca | 2001-09-14 16:35:16 +0000 | [diff] [blame] | 1494 | |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1495 | - The "compiler" package has been brought up to date to the state of |
Guido van Rossum | c9ed5dc | 2001-09-20 05:30:24 +0000 | [diff] [blame] | 1496 | Python 2.2 bytecode generation. It has also been promoted from a |
| 1497 | Tool to a standard library package. (Tools/compiler still exists as |
| 1498 | a sample driver.) |
| 1499 | |
Guido van Rossum | c9ed5dc | 2001-09-20 05:30:24 +0000 | [diff] [blame] | 1500 | Tools |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1501 | ----- |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1502 | |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1503 | Build |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1504 | ----- |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1505 | |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1506 | - Large file support (LFS) is now automatic when the platform supports |
| 1507 | it; no more manual configuration tweaks are needed. On Linux, at |
| 1508 | least, it's possible to have a system whose C library supports large |
| 1509 | files but whose kernel doesn't; in this case, large file support is |
| 1510 | still enabled but doesn't do you any good unless you upgrade your |
| 1511 | kernel or share your Python executable with another system whose |
| 1512 | kernel has large file support. |
| 1513 | |
| 1514 | - The configure script now supplies plausible defaults in a |
| 1515 | cross-compilation environment. This doesn't mean that the supplied |
| 1516 | values are always correct, or that cross-compilation now works |
| 1517 | flawlessly -- but it's a first step (and it shuts up most of |
| 1518 | autoconf's warnings about AC_TRY_RUN). |
| 1519 | |
| 1520 | - The Unix build is now a bit less chatty, courtesy of the parser |
| 1521 | generator. The build is completely silent (except for errors) when |
| 1522 | using "make -s", thanks to a -q option to setup.py. |
| 1523 | |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1524 | C API |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1525 | ----- |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1526 | |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1527 | - The "structmember" API now supports some new flag bits to deny read |
| 1528 | and/or write access to attributes in restricted execution mode. |
| 1529 | |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1530 | New platforms |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1531 | ------------- |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1532 | |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1533 | - Compaq's iPAQ handheld, running the "familiar" Linux distribution |
| 1534 | (http://familiar.handhelds.org). |
| 1535 | |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1536 | Tests |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1537 | ----- |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1538 | |
Tim Peters | 8a9c284 | 2001-09-22 21:30:22 +0000 | [diff] [blame] | 1539 | - The "classic" standard tests, which work by comparing stdout to |
| 1540 | an expected-output file under Lib/test/output/, no longer stop at |
| 1541 | the first mismatch. Instead the test is run to completion, and a |
| 1542 | variant of ndiff-style comparison is used to report all differences. |
| 1543 | This is much easier to understand than the previous style of reporting. |
| 1544 | |
| 1545 | - The unittest-based standard tests now use regrtest's test_main() |
| 1546 | convention, instead of running as a side-effect of merely being |
| 1547 | imported. This allows these tests to be run in more natural and |
| 1548 | flexible ways as unittests, outside the regrtest framework. |
| 1549 | |
| 1550 | - regrtest.py is much better integrated with unittest and doctest now, |
| 1551 | especially in regard to reporting errors. |
| 1552 | |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1553 | Windows |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1554 | ------- |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1555 | |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1556 | - Large file support now also works for files > 4GB, on filesystems |
Tim Peters | 8a9c284 | 2001-09-22 21:30:22 +0000 | [diff] [blame] | 1557 | that support it (NTFS under Windows 2000). See "What's New in |
| 1558 | Python 2.2a3" for more detail. |
Guido van Rossum | 624c8af | 2001-09-18 15:21:04 +0000 | [diff] [blame] | 1559 | |
Tim Peters | b07352e | 2001-09-08 01:25:47 +0000 | [diff] [blame] | 1560 | |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1561 | What's New in Python 2.2a3? |
| 1562 | =========================== |
| 1563 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1564 | *Release Date: 07-Sep-2001* |
| 1565 | |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1566 | Core |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1567 | ---- |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1568 | |
Tim Peters | 9fffa3e | 2001-09-04 05:14:19 +0000 | [diff] [blame] | 1569 | - Conversion of long to float now raises OverflowError if the long is too |
| 1570 | big to represent as a C double. |
| 1571 | |
Tim Peters | 32f453e | 2001-09-03 08:35:41 +0000 | [diff] [blame] | 1572 | - The 3-argument builtin pow() no longer allows a third non-None argument |
| 1573 | if either of the first two arguments is a float, or if both are of |
| 1574 | integer types and the second argument is negative (in which latter case |
| 1575 | the arguments are converted to float, so this is really the same |
| 1576 | restriction). |
| 1577 | |
Tim Peters | 5d2b77c | 2001-09-03 05:47:38 +0000 | [diff] [blame] | 1578 | - The builtin dir() now returns more information, and sometimes much |
| 1579 | more, generally naming all attributes of an object, and all attributes |
| 1580 | reachable from the object via its class, and from its class's base |
| 1581 | classes, and so on from them too. Example: in 2.2a2, dir([]) returned |
| 1582 | an empty list. In 2.2a3, |
| 1583 | |
| 1584 | >>> dir([]) |
| 1585 | ['__add__', '__class__', '__contains__', '__delattr__', '__delitem__', |
| 1586 | '__eq__', '__ge__', '__getattr__', '__getitem__', '__getslice__', |
| 1587 | '__gt__', '__hash__', '__iadd__', '__imul__', '__init__', '__le__', |
| 1588 | '__len__', '__lt__', '__mul__', '__ne__', '__new__', '__repr__', |
| 1589 | '__rmul__', '__setattr__', '__setitem__', '__setslice__', '__str__', |
| 1590 | 'append', 'count', 'extend', 'index', 'insert', 'pop', 'remove', |
| 1591 | 'reverse', 'sort'] |
| 1592 | |
| 1593 | dir(module) continues to return only the module's attributes, though. |
| 1594 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1595 | - Overflowing operations on plain ints now return a long int rather |
Guido van Rossum | ae45714 | 2001-08-31 18:31:35 +0000 | [diff] [blame] | 1596 | than raising OverflowError. This is a partial implementation of PEP |
| 1597 | 237. You can use -Wdefault::OverflowWarning to enable a warning for |
| 1598 | this situation, and -Werror::OverflowWarning to revert to the old |
| 1599 | OverflowError exception. |
| 1600 | |
Guido van Rossum | 61c345f | 2001-09-04 03:26:15 +0000 | [diff] [blame] | 1601 | - A new command line option, -Q<arg>, is added to control run-time |
Guido van Rossum | ae45714 | 2001-08-31 18:31:35 +0000 | [diff] [blame] | 1602 | warnings for the use of classic division. (See PEP 238.) Possible |
Barry Warsaw | d6c8ca6 | 2001-09-07 18:13:44 +0000 | [diff] [blame] | 1603 | values are -Qold, -Qwarn, -Qwarnall, and -Qnew. The default is |
| 1604 | -Qold, meaning the / operator has its classic meaning and no |
| 1605 | warnings are issued. Using -Qwarn issues a run-time warning about |
| 1606 | all uses of classic division for int and long arguments; -Qwarnall |
| 1607 | also warns about classic division for float and complex arguments |
Tim Peters | 3caca23 | 2001-12-06 06:23:26 +0000 | [diff] [blame] | 1608 | (for use with fixdiv.py). |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1609 | [Note: the remainder of this item (preserved below) became |
| 1610 | obsolete in 2.2c1 -- -Qnew has global effect in 2.2] :: |
| 1611 | |
| 1612 | Using -Qnew is questionable; it turns on new division by default, but |
| 1613 | only in the __main__ module. You can usefully combine -Qwarn or |
| 1614 | -Qwarnall and -Qnew: this gives the __main__ module new division, and |
| 1615 | warns about classic division everywhere else. |
Guido van Rossum | ae45714 | 2001-08-31 18:31:35 +0000 | [diff] [blame] | 1616 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1617 | - Many built-in types can now be subclassed. This applies to int, |
Guido van Rossum | ae45714 | 2001-08-31 18:31:35 +0000 | [diff] [blame] | 1618 | long, float, str, unicode, and tuple. (The types complex, list and |
| 1619 | dictionary can also be subclassed; this was introduced earlier.) |
| 1620 | Note that restrictions apply when subclassing immutable built-in |
| 1621 | types: you can only affect the value of the instance by overloading |
| 1622 | __new__. You can add mutable attributes, and the subclass instances |
| 1623 | will have a __dict__ attribute, but you cannot change the "value" |
| 1624 | (as implemented by the base class) of an immutable subclass instance |
| 1625 | once it is created. |
| 1626 | |
Guido van Rossum | aaf80c8 | 2001-09-02 13:44:35 +0000 | [diff] [blame] | 1627 | - The dictionary constructor now takes an optional argument, a |
| 1628 | mapping-like object, and initializes the dictionary from its |
| 1629 | (key, value) pairs. |
| 1630 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1631 | - A new built-in type, super, has been added. This facilitates making |
Guido van Rossum | ae45714 | 2001-08-31 18:31:35 +0000 | [diff] [blame] | 1632 | "cooperative super calls" in a multiple inheritance setting. For an |
| 1633 | explanation, see http://www.python.org/2.2/descrintro.html#cooperation |
| 1634 | |
Guido van Rossum | 8d7234d | 2001-09-06 22:02:58 +0000 | [diff] [blame] | 1635 | - A new built-in type, property, has been added. This enables the |
| 1636 | creation of "properties". These are attributes implemented by |
| 1637 | getter and setter functions (or only one of these for read-only or |
| 1638 | write-only attributes), without the need to override __getattr__. |
| 1639 | See http://www.python.org/2.2/descrintro.html#property |
Guido van Rossum | ae45714 | 2001-08-31 18:31:35 +0000 | [diff] [blame] | 1640 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1641 | - The syntax of floating-point and imaginary literals has been |
Tim Peters | d507dab | 2001-08-30 20:51:59 +0000 | [diff] [blame] | 1642 | liberalized, to allow leading zeroes. Examples of literals now |
| 1643 | legal that were SyntaxErrors before: |
| 1644 | |
| 1645 | 00.0 0e3 0100j 07.5 00000000000000000008. |
| 1646 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1647 | - An old tokenizer bug allowed floating point literals with an incomplete |
Tim Peters | c6d9581 | 2001-08-28 20:56:27 +0000 | [diff] [blame] | 1648 | exponent, such as 1e and 3.1e-. Such literals now raise SyntaxError. |
| 1649 | |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1650 | Library |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1651 | ------- |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1652 | |
Martin v. Löwis | cb227c9 | 2001-09-06 08:54:16 +0000 | [diff] [blame] | 1653 | - telnetlib includes symbolic names for the options, and support for |
| 1654 | setting an option negotiation callback. |
| 1655 | |
Tim Peters | a40c793 | 2001-09-05 22:36:56 +0000 | [diff] [blame] | 1656 | - The new C standard no longer requires that math libraries set errno to |
| 1657 | ERANGE on overflow. For platform libraries that exploit this new |
| 1658 | freedom, Python's overflow-checking was wholly broken. A new overflow- |
| 1659 | checking scheme attempts to repair that, but may not be reliable on all |
| 1660 | platforms (C doesn't seem to provide anything both useful and portable |
| 1661 | in this area anymore). |
| 1662 | |
Martin v. Löwis | 44f8696 | 2001-09-05 13:44:54 +0000 | [diff] [blame] | 1663 | - Asynchronous timeout actions are available through the new class |
| 1664 | threading.Timer. |
| 1665 | |
Tim Peters | 7852616 | 2001-09-05 00:53:45 +0000 | [diff] [blame] | 1666 | - math.log and math.log10 now return sensible results for even huge |
| 1667 | long arguments. For example, math.log10(10 ** 10000) ~= 10000.0. |
| 1668 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1669 | - A new function, imp.lock_held(), returns 1 when the import lock is |
Tim Peters | 6923234 | 2001-08-30 05:16:13 +0000 | [diff] [blame] | 1670 | currently held. See the docs for the imp module. |
| 1671 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1672 | - pickle, cPickle and marshal on 32-bit platforms can now correctly read |
Tim Peters | 8211237 | 2001-08-29 02:28:42 +0000 | [diff] [blame] | 1673 | dumps containing ints written on platforms where Python ints are 8 bytes. |
| 1674 | When read on a box where Python ints are 4 bytes, such values are |
| 1675 | converted to Python longs. |
| 1676 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1677 | - In restricted execution mode (using the rexec module), unmarshalling |
Guido van Rossum | ae45714 | 2001-08-31 18:31:35 +0000 | [diff] [blame] | 1678 | code objects is no longer allowed. This plugs a security hole. |
| 1679 | |
Steve Purcell | 6091cd6 | 2001-09-06 16:05:17 +0000 | [diff] [blame] | 1680 | - unittest.TestResult instances no longer store references to tracebacks |
| 1681 | generated by test failures. This prevents unexpected dangling references |
| 1682 | to objects that should be garbage collected between tests. |
| 1683 | |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1684 | Tools |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1685 | ----- |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1686 | |
Barry Warsaw | d6c8ca6 | 2001-09-07 18:13:44 +0000 | [diff] [blame] | 1687 | - Tools/scripts/fixdiv.py has been added which can be used to fix |
| 1688 | division operators as per PEP 238. |
| 1689 | |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1690 | Build |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1691 | ----- |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1692 | |
Barry Warsaw | d6c8ca6 | 2001-09-07 18:13:44 +0000 | [diff] [blame] | 1693 | - If you are an adventurous person using Mac OS X you may want to look at |
| 1694 | Mac/OSX. There is a Makefile there that will build Python as a real Mac |
| 1695 | application, which can be used for experimenting with Carbon or Cocoa. |
| 1696 | Discussion of this on pythonmac-sig, please. |
| 1697 | |
Tim Peters | 7eea37e | 2001-09-04 22:08:56 +0000 | [diff] [blame] | 1698 | C API |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1699 | ----- |
Tim Peters | 7eea37e | 2001-09-04 22:08:56 +0000 | [diff] [blame] | 1700 | |
| 1701 | - New function PyObject_Dir(obj), like Python __builtin__.dir(obj). |
Tim Peters | 6923234 | 2001-08-30 05:16:13 +0000 | [diff] [blame] | 1702 | |
Tim Peters | 9fffa3e | 2001-09-04 05:14:19 +0000 | [diff] [blame] | 1703 | - Note that PyLong_AsDouble can fail! This has always been true, but no |
| 1704 | callers checked for it. It's more likely to fail now, because overflow |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1705 | errors are properly detected now. The proper way to check: :: |
Tim Peters | 9fffa3e | 2001-09-04 05:14:19 +0000 | [diff] [blame] | 1706 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1707 | double x = PyLong_AsDouble(some_long_object); |
| 1708 | if (x == -1.0 && PyErr_Occurred()) { |
| 1709 | /* The conversion failed. */ |
| 1710 | } |
Tim Peters | 9fffa3e | 2001-09-04 05:14:19 +0000 | [diff] [blame] | 1711 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1712 | - The GC API has been changed. Extensions that use the old API will still |
Neil Schemenauer | 4042c69 | 2001-08-30 15:38:01 +0000 | [diff] [blame] | 1713 | compile but will not participate in GC. To upgrade an extension |
| 1714 | module: |
| 1715 | |
| 1716 | - rename Py_TPFLAGS_GC to PyTPFLAGS_HAVE_GC |
Tim Peters | d507dab | 2001-08-30 20:51:59 +0000 | [diff] [blame] | 1717 | |
Neil Schemenauer | 4042c69 | 2001-08-30 15:38:01 +0000 | [diff] [blame] | 1718 | - use PyObject_GC_New or PyObject_GC_NewVar to allocate objects and |
| 1719 | PyObject_GC_Del to deallocate them |
Tim Peters | d507dab | 2001-08-30 20:51:59 +0000 | [diff] [blame] | 1720 | |
Neil Schemenauer | 4042c69 | 2001-08-30 15:38:01 +0000 | [diff] [blame] | 1721 | - rename PyObject_GC_Init to PyObject_GC_Track and PyObject_GC_Fini |
| 1722 | to PyObject_GC_UnTrack |
Tim Peters | d507dab | 2001-08-30 20:51:59 +0000 | [diff] [blame] | 1723 | |
Neil Schemenauer | 4042c69 | 2001-08-30 15:38:01 +0000 | [diff] [blame] | 1724 | - remove PyGC_HEAD_SIZE from object size calculations |
| 1725 | |
| 1726 | - remove calls to PyObject_AS_GC and PyObject_FROM_GC |
| 1727 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1728 | - Two new functions: PyString_FromFormat() and PyString_FromFormatV(). |
Guido van Rossum | ae45714 | 2001-08-31 18:31:35 +0000 | [diff] [blame] | 1729 | These can be used safely to construct string objects from a |
| 1730 | sprintf-style format string (similar to the format string supported |
| 1731 | by PyErr_Format()). |
Tim Peters | 6923234 | 2001-08-30 05:16:13 +0000 | [diff] [blame] | 1732 | |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1733 | New platforms |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1734 | ------------- |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1735 | |
Tim Peters | b7da090 | 2001-09-02 23:01:43 +0000 | [diff] [blame] | 1736 | - Stephen Hansen contributed patches sufficient to get a clean compile |
| 1737 | under Borland C (Windows), but he reports problems running it and ran |
| 1738 | out of time to complete the port. Volunteers? Expect a MemoryError |
| 1739 | when importing the types module; this is probably shallow, and |
| 1740 | causing later failures too. |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1741 | |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1742 | Tests |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1743 | ----- |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1744 | |
| 1745 | Windows |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1746 | ------- |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1747 | |
Tim Peters | 6e13a56 | 2001-09-06 00:32:15 +0000 | [diff] [blame] | 1748 | - Large file support is now enabled on Win32 platforms as well as on |
| 1749 | Win64. This means that, for example, you can use f.tell() and f.seek() |
| 1750 | to manipulate files larger than 2 gigabytes (provided you have enough |
| 1751 | disk space, and are using a Windows filesystem that supports large |
Tim Peters | 9a9471c | 2001-09-11 23:18:51 +0000 | [diff] [blame] | 1752 | partitions). Windows filesystem limits: FAT has a 2GB (gigabyte) |
| 1753 | filesize limit, and large file support makes no difference there. |
| 1754 | FAT32's limit is 4GB, and files >= 2GB are easier to use from Python now. |
| 1755 | NTFS has no practical limit on file size, and files of any size can be |
| 1756 | used from Python now. |
Tim Peters | 6e13a56 | 2001-09-06 00:32:15 +0000 | [diff] [blame] | 1757 | |
Tim Peters | bdee63f | 2001-09-02 03:40:59 +0000 | [diff] [blame] | 1758 | - The w9xpopen hack is now used on Windows NT and 2000 too when COMPSPEC |
Tim Peters | 402d598 | 2001-08-27 06:37:48 +0000 | [diff] [blame] | 1759 | points to command.com (patch from Brian Quinlan). |
| 1760 | |
Tim Peters | edc9931 | 2001-08-22 21:36:50 +0000 | [diff] [blame] | 1761 | |
Tim Peters | 20f51a7 | 2001-07-21 02:31:40 +0000 | [diff] [blame] | 1762 | What's New in Python 2.2a2? |
| 1763 | =========================== |
| 1764 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1765 | *Release Date: 22-Aug-2001* |
| 1766 | |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 1767 | Build |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1768 | ----- |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 1769 | |
Barry Warsaw | 18b2ecf | 2001-08-22 20:26:56 +0000 | [diff] [blame] | 1770 | - Tim Peters developed a brand new Windows installer using Wise 8.1, |
| 1771 | generously donated to us by Wise Solutions. |
| 1772 | |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 1773 | - configure supports a new option --enable-unicode, with the values |
| 1774 | ucs2 and ucs4 (new in 2.2a1). With --disable-unicode, the Unicode |
| 1775 | type and supporting code is completely removed from the interpreter. |
Barry Warsaw | 6f3410d | 2001-08-22 04:08:41 +0000 | [diff] [blame] | 1776 | |
Jack Jansen | 32ce0cd | 2001-08-21 19:28:20 +0000 | [diff] [blame] | 1777 | - A new configure option --enable-framework builds a Mac OS X framework, |
| 1778 | which "make frameworkinstall" will install. This provides a starting |
| 1779 | point for more mac-like functionality, join pythonmac-sig@python.org |
| 1780 | if you are interested in helping. |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 1781 | |
Barry Warsaw | 6f3410d | 2001-08-22 04:08:41 +0000 | [diff] [blame] | 1782 | - The NeXT platform is no longer supported. |
| 1783 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1784 | - The 'new' module is now statically linked. |
Barry Warsaw | 6f3410d | 2001-08-22 04:08:41 +0000 | [diff] [blame] | 1785 | |
Tim Peters | 0afb609 | 2001-08-15 06:06:44 +0000 | [diff] [blame] | 1786 | Tools |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1787 | ----- |
Tim Peters | 0afb609 | 2001-08-15 06:06:44 +0000 | [diff] [blame] | 1788 | |
| 1789 | - The new Tools/scripts/cleanfuture.py can be used to automatically |
Andrew M. Kuchling | b053209 | 2001-08-15 15:54:56 +0000 | [diff] [blame] | 1790 | edit out obsolete future statements from Python source code. See |
Tim Peters | 0afb609 | 2001-08-15 06:06:44 +0000 | [diff] [blame] | 1791 | the module docstring for details. |
| 1792 | |
Tim Peters | 20f51a7 | 2001-07-21 02:31:40 +0000 | [diff] [blame] | 1793 | Tests |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1794 | ----- |
Tim Peters | 20f51a7 | 2001-07-21 02:31:40 +0000 | [diff] [blame] | 1795 | |
Tim Peters | 5e824c3 | 2001-08-12 22:25:01 +0000 | [diff] [blame] | 1796 | - regrtest.py now knows which tests are expected to be skipped on some |
Barry Warsaw | 6f3410d | 2001-08-22 04:08:41 +0000 | [diff] [blame] | 1797 | platforms, allowing to give clearer test result output. regrtest |
| 1798 | also has optional --use/-u switch to run normally disabled tests |
| 1799 | which require network access or consume significant disk resources. |
Tim Peters | 5e824c3 | 2001-08-12 22:25:01 +0000 | [diff] [blame] | 1800 | |
Tim Peters | 20f51a7 | 2001-07-21 02:31:40 +0000 | [diff] [blame] | 1801 | - Several new tests in the standard test suite, with special thanks to |
| 1802 | Nick Mathewson. |
| 1803 | |
| 1804 | Core |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1805 | ---- |
Tim Peters | 20f51a7 | 2001-07-21 02:31:40 +0000 | [diff] [blame] | 1806 | |
Barry Warsaw | 6f3410d | 2001-08-22 04:08:41 +0000 | [diff] [blame] | 1807 | - The floor division operator // has been added as outlined in PEP |
| 1808 | 238. The / operator still provides classic division (and will until |
| 1809 | Python 3.0) unless "from __future__ import division" is included, in |
| 1810 | which case the / operator will provide true division. The operator |
| 1811 | module provides truediv() and floordiv() functions. Augmented |
| 1812 | assignment variants are included, as are the equivalent overloadable |
| 1813 | methods and C API methods. See the PEP for a full discussion: |
| 1814 | <http://python.sf.net/peps/pep-0238.html> |
| 1815 | |
Tim Peters | 6cd6a82 | 2001-08-17 22:11:27 +0000 | [diff] [blame] | 1816 | - Future statements are now effective in simulated interactive shells |
| 1817 | (like IDLE). This should "just work" by magic, but read Michael |
| 1818 | Hudson's "Future statements in simulated shells" PEP 264 for full |
| 1819 | details: <http://python.sf.net/peps/pep-0264.html>. |
| 1820 | |
Guido van Rossum | f86ddd2 | 2001-08-17 21:21:04 +0000 | [diff] [blame] | 1821 | - The type/class unification (PEP 252-253) was integrated into the |
| 1822 | trunk and is not so tentative any more (the exact specification of |
| 1823 | some features is still tentative). A lot of work has done on fixing |
| 1824 | bugs and adding robustness and features (performance still has to |
| 1825 | come a long way). |
| 1826 | |
Marc-André Lemburg | 888fac0 | 2001-07-31 14:24:31 +0000 | [diff] [blame] | 1827 | - Warnings about a mismatch in the Python API during extension import |
| 1828 | now use the Python warning framework (which makes it possible to |
| 1829 | write filters for these warnings). |
| 1830 | |
Barry Warsaw | 9b3be7f | 2001-08-14 18:35:02 +0000 | [diff] [blame] | 1831 | - A function's __dict__ (aka func_dict) will now always be a |
| 1832 | dictionary. It used to be possible to delete it or set it to None, |
| 1833 | but now both actions raise TypeErrors. It is still legal to set it |
| 1834 | to a dictionary object. Getting func.__dict__ before any attributes |
| 1835 | have been assigned now returns an empty dictionary instead of None. |
| 1836 | |
Guido van Rossum | 32aa5d2 | 2001-09-05 18:43:35 +0000 | [diff] [blame] | 1837 | - A new command line option, -E, was added which disables the use of |
| 1838 | all environment variables, or at least those that are specifically |
| 1839 | significant to Python. Usually those have a name starting with |
| 1840 | "PYTHON". This was used to fix a problem where the tests fail if |
| 1841 | the user happens to have PYTHONHOME or PYTHONPATH pointing to an |
| 1842 | older distribution. |
| 1843 | |
Marc-André Lemburg | d627791 | 2001-07-31 14:42:42 +0000 | [diff] [blame] | 1844 | Library |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1845 | ------- |
Marc-André Lemburg | d627791 | 2001-07-31 14:42:42 +0000 | [diff] [blame] | 1846 | |
Tim Peters | 5e824c3 | 2001-08-12 22:25:01 +0000 | [diff] [blame] | 1847 | - New class Differ and new functions ndiff() and restore() in difflib.py. |
| 1848 | These package the algorithms used by the popular Tools/scripts/ndiff.py, |
Tim Peters | 0afb609 | 2001-08-15 06:06:44 +0000 | [diff] [blame] | 1849 | for programmatic reuse. |
Tim Peters | 5e824c3 | 2001-08-12 22:25:01 +0000 | [diff] [blame] | 1850 | |
Marc-André Lemburg | d627791 | 2001-07-31 14:42:42 +0000 | [diff] [blame] | 1851 | - New function xml.sax.saxutils.quoteattr(): Quote an XML attribute |
| 1852 | value using the minimal quoting required for the value; more |
| 1853 | reliable than using xml.sax.saxutils.escape() for attribute values. |
| 1854 | |
| 1855 | - Readline completion support for cmd.Cmd was added. |
| 1856 | |
Barry Warsaw | 6f3410d | 2001-08-22 04:08:41 +0000 | [diff] [blame] | 1857 | - Calling os.tempnam() or os.tmpnam() generate RuntimeWarnings. |
| 1858 | |
| 1859 | - Added function threading.BoundedSemaphore() |
| 1860 | |
| 1861 | - Added Ka-Ping Yee's cgitb.py module. |
| 1862 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1863 | - The 'new' module now exposes the CO_xxx flags. |
Barry Warsaw | 6f3410d | 2001-08-22 04:08:41 +0000 | [diff] [blame] | 1864 | |
Martin v. Löwis | 560da62 | 2001-11-24 09:24:51 +0000 | [diff] [blame] | 1865 | - The gc module offers the get_referents function. |
| 1866 | |
Marc-André Lemburg | d627791 | 2001-07-31 14:42:42 +0000 | [diff] [blame] | 1867 | New platforms |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1868 | ------------- |
Marc-André Lemburg | d627791 | 2001-07-31 14:42:42 +0000 | [diff] [blame] | 1869 | |
| 1870 | C API |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1871 | ----- |
Marc-André Lemburg | d627791 | 2001-07-31 14:42:42 +0000 | [diff] [blame] | 1872 | |
Marc-André Lemburg | 888fac0 | 2001-07-31 14:24:31 +0000 | [diff] [blame] | 1873 | - Two new APIs PyOS_snprintf() and PyOS_vsnprintf() were added |
| 1874 | which provide a cross-platform implementations for the |
| 1875 | relatively new snprintf()/vsnprintf() C lib APIs. In contrast to |
| 1876 | the standard sprintf() and vsprintf() C lib APIs, these versions |
| 1877 | apply bounds checking on the used buffer which enhances protection |
| 1878 | against buffer overruns. |
| 1879 | |
Marc-André Lemburg | 48dbfe9 | 2001-07-31 14:37:40 +0000 | [diff] [blame] | 1880 | - Unicode APIs now use name mangling to assure that mixing interpreters |
Tim Peters | c173137 | 2001-08-04 08:12:36 +0000 | [diff] [blame] | 1881 | and extensions using different Unicode widths is rendered next to |
| 1882 | impossible. Trying to import an incompatible Unicode-aware extension |
Marc-André Lemburg | 48dbfe9 | 2001-07-31 14:37:40 +0000 | [diff] [blame] | 1883 | will result in an ImportError. Unicode extensions writers must make |
| 1884 | sure to check the Unicode width compatibility in their extensions by |
| 1885 | using at least one of the mangled Unicode APIs in the extension. |
| 1886 | |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 1887 | - Two new flags METH_NOARGS and METH_O are available in method definition |
| 1888 | tables to simplify implementation of methods with no arguments and a |
| 1889 | single untyped argument. Calling such methods is more efficient than |
| 1890 | calling corresponding METH_VARARGS methods. METH_OLDARGS is now |
| 1891 | deprecated. |
| 1892 | |
Tim Peters | c173137 | 2001-08-04 08:12:36 +0000 | [diff] [blame] | 1893 | Windows |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1894 | ------- |
Tim Peters | c173137 | 2001-08-04 08:12:36 +0000 | [diff] [blame] | 1895 | |
| 1896 | - "import module" now compiles module.pyw if it exists and nothing else |
| 1897 | relevant is found. |
| 1898 | |
Tim Peters | 20f51a7 | 2001-07-21 02:31:40 +0000 | [diff] [blame] | 1899 | |
Guido van Rossum | 55a7899 | 2001-07-17 17:22:32 +0000 | [diff] [blame] | 1900 | What's New in Python 2.2a1? |
Tim Peters | f553f89 | 2001-05-01 20:45:31 +0000 | [diff] [blame] | 1901 | =========================== |
| 1902 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1903 | *Release date: 18-Jul-2001* |
| 1904 | |
Tim Peters | f553f89 | 2001-05-01 20:45:31 +0000 | [diff] [blame] | 1905 | Core |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1906 | ---- |
Tim Peters | a814db5 | 2001-05-14 07:05:58 +0000 | [diff] [blame] | 1907 | |
Tim Peters | 6d6c1a3 | 2001-08-02 04:15:00 +0000 | [diff] [blame] | 1908 | - TENTATIVELY, a large amount of code implementing much of what's |
| 1909 | described in PEP 252 (Making Types Look More Like Classes) and PEP |
| 1910 | 253 (Subtyping Built-in Types) was added. This will be released |
| 1911 | with Python 2.2a1. Documentation will be provided separately |
| 1912 | through http://www.python.org/2.2/. The purpose of releasing this |
| 1913 | with Python 2.2a1 is to test backwards compatibility. It is |
| 1914 | possible, though not likely, that a decision is made not to release |
| 1915 | this code as part of 2.2 final, if any serious backwards |
Raymond Hettinger | f2f2a2c | 2002-08-19 14:25:03 +0000 | [diff] [blame] | 1916 | incompatibilities are found during alpha testing that cannot be |
Tim Peters | 6d6c1a3 | 2001-08-02 04:15:00 +0000 | [diff] [blame] | 1917 | repaired. |
| 1918 | |
Guido van Rossum | 55a7899 | 2001-07-17 17:22:32 +0000 | [diff] [blame] | 1919 | - Generators were added; this is a new way to create an iterator (see |
Tim Peters | 11a0d10 | 2001-07-17 18:48:00 +0000 | [diff] [blame] | 1920 | below) using what looks like a simple function containing one or |
Guido van Rossum | 55a7899 | 2001-07-17 17:22:32 +0000 | [diff] [blame] | 1921 | more 'yield' statements. See PEP 255. Since this adds a new |
| 1922 | keyword to the language, this feature must be enabled by including a |
| 1923 | future statement: "from __future__ import generators" (see PEP 236). |
| 1924 | Generators will become a standard feature in a future release |
| 1925 | (probably 2.3). Without this future statement, 'yield' remains an |
| 1926 | ordinary identifier, but a warning is issued each time it is used. |
| 1927 | (These warnings currently don't conform to the warnings framework of |
| 1928 | PEP 230; we intend to fix this in 2.2a2.) |
| 1929 | |
Marc-André Lemburg | 12e74b3 | 2001-05-22 08:58:23 +0000 | [diff] [blame] | 1930 | - The UTF-16 codec was modified to be more RFC compliant. It will now |
| 1931 | only remove BOM characters at the start of the string and then |
| 1932 | only if running in native mode (UTF-16-LE and -BE won't remove a |
| 1933 | leading BMO character). |
| 1934 | |
Marc-André Lemburg | fab96cc | 2001-05-15 18:38:45 +0000 | [diff] [blame] | 1935 | - Strings now have a new method .decode() to complement the already |
| 1936 | existing .encode() method. These two methods provide direct access |
| 1937 | to the corresponding decoders and encoders of the registered codecs. |
| 1938 | |
| 1939 | To enhance the usability of the .encode() method, the special |
| 1940 | casing of Unicode object return values was dropped (Unicode objects |
| 1941 | were auto-magically converted to string using the default encoding). |
Tim Peters | eb28ef2 | 2001-06-02 05:27:19 +0000 | [diff] [blame] | 1942 | |
Marc-André Lemburg | fab96cc | 2001-05-15 18:38:45 +0000 | [diff] [blame] | 1943 | Both methods will now return whatever the codec in charge of the |
| 1944 | requested encoding returns as object, e.g. Unicode codecs will |
| 1945 | return Unicode objects when decoding is requested ("äöü".decode("latin-1") |
| 1946 | will return u"äöü"). This enables codec writer to create codecs |
| 1947 | for various simple to use conversions. |
| 1948 | |
| 1949 | New codecs were added to demonstrate these new features (the .encode() |
| 1950 | and .decode() columns indicate the type of the returned objects): |
| 1951 | |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 1952 | +---------+-----------+-----------+-----------------------------+ |
| 1953 | |Name | .encode() | .decode() | Description | |
| 1954 | +=========+===========+===========+=============================+ |
| 1955 | |uu | string | string | UU codec (e.g. for email) | |
| 1956 | +---------+-----------+-----------+-----------------------------+ |
| 1957 | |base64 | string | string | base64 codec | |
| 1958 | +---------+-----------+-----------+-----------------------------+ |
| 1959 | |quopri | string | string | quoted-printable codec | |
| 1960 | +---------+-----------+-----------+-----------------------------+ |
| 1961 | |zlib | string | string | zlib compression | |
| 1962 | +---------+-----------+-----------+-----------------------------+ |
| 1963 | |hex | string | string | 2-byte hex codec | |
| 1964 | +---------+-----------+-----------+-----------------------------+ |
| 1965 | |rot-13 | string | Unicode | ROT-13 Unicode charmap codec| |
| 1966 | +---------+-----------+-----------+-----------------------------+ |
Marc-André Lemburg | fab96cc | 2001-05-15 18:38:45 +0000 | [diff] [blame] | 1967 | |
Mark Hammond | 2a0af79 | 2001-05-14 03:09:36 +0000 | [diff] [blame] | 1968 | - Some operating systems now support the concept of a default Unicode |
| 1969 | encoding for file system operations. Notably, Windows supports 'mbcs' |
| 1970 | as the default. The Macintosh will also adopt this concept in the medium |
Guido van Rossum | 1bd797a | 2001-05-14 13:53:38 +0000 | [diff] [blame] | 1971 | term, although the default encoding for that platform will be other than |
Mark Hammond | 2a0af79 | 2001-05-14 03:09:36 +0000 | [diff] [blame] | 1972 | 'mbcs'. |
Guido van Rossum | 1bd797a | 2001-05-14 13:53:38 +0000 | [diff] [blame] | 1973 | |
| 1974 | On operating system that support non-ASCII filenames, it is common for |
Mark Hammond | 2a0af79 | 2001-05-14 03:09:36 +0000 | [diff] [blame] | 1975 | functions that return filenames (such as os.listdir()) to return Python |
| 1976 | string objects pre-encoded using the default file system encoding for |
| 1977 | the platform. As this encoding is likely to be different from Python's |
| 1978 | default encoding, converting this name to a Unicode object before passing |
| 1979 | it back to the Operating System would result in a Unicode error, as Python |
Tim Peters | a814db5 | 2001-05-14 07:05:58 +0000 | [diff] [blame] | 1980 | would attempt to use its default encoding (generally ASCII) rather than |
| 1981 | the default encoding for the file system. |
Guido van Rossum | 1bd797a | 2001-05-14 13:53:38 +0000 | [diff] [blame] | 1982 | |
Tim Peters | a814db5 | 2001-05-14 07:05:58 +0000 | [diff] [blame] | 1983 | In general, this change simply removes surprises when working with |
| 1984 | Unicode and the file system, making these operations work as you expect, |
| 1985 | increasing the transparency of Unicode objects in this context. |
Mark Hammond | 2a0af79 | 2001-05-14 03:09:36 +0000 | [diff] [blame] | 1986 | See [????] for more details, including examples. |
Tim Peters | f553f89 | 2001-05-01 20:45:31 +0000 | [diff] [blame] | 1987 | |
Tim Peters | 61dff2b | 2001-05-08 15:43:37 +0000 | [diff] [blame] | 1988 | - Float (and complex) literals in source code were evaluated to full |
| 1989 | precision only when running from a .py file; the same code loaded from a |
| 1990 | .pyc (or .pyo) file could suffer numeric differences starting at about the |
| 1991 | 12th significant decimal digit. For example, on a machine with IEEE-754 |
| 1992 | floating arithmetic, |
| 1993 | |
| 1994 | x = 9007199254740992.0 |
| 1995 | print long(x) |
| 1996 | |
| 1997 | printed 9007199254740992 if run directly from .py, but 9007199254740000 |
| 1998 | if from a compiled (.pyc or .pyo) file. This was due to marshal using |
| 1999 | str(float) instead of repr(float) when building code objects. marshal |
| 2000 | now uses repr(float) instead, which should reproduce floats to full |
| 2001 | machine precision (assuming the platform C float<->string I/O conversion |
| 2002 | functions are of good quality). |
| 2003 | |
| 2004 | This may cause floating-point results to change in some cases, and |
| 2005 | usually for the better, but may also cause numerically unstable |
| 2006 | algorithms to break. |
| 2007 | |
Tim Peters | 2f228e7 | 2001-05-13 00:19:31 +0000 | [diff] [blame] | 2008 | - The implementation of dicts suffers fewer collisions, which has speed |
| 2009 | benefits. However, the order in which dict entries appear in dict.keys(), |
| 2010 | dict.values() and dict.items() may differ from previous releases for a |
| 2011 | given dict. Nothing is defined about this order, so no program should |
| 2012 | rely on it. Nevertheless, it's easy to write test cases that rely on the |
| 2013 | order by accident, typically because of printing the str() or repr() of a |
| 2014 | dict to an "expected results" file. See Lib/test/test_support.py's new |
| 2015 | sortdict(dict) function for a simple way to display a dict in sorted |
| 2016 | order. |
| 2017 | |
Tim Peters | 7a3bfc3 | 2001-06-12 01:22:22 +0000 | [diff] [blame] | 2018 | - Many other small changes to dicts were made, resulting in faster |
| 2019 | operation along the most common code paths. |
| 2020 | |
Guido van Rossum | 1031582 | 2001-05-01 20:54:30 +0000 | [diff] [blame] | 2021 | - Dictionary objects now support the "in" operator: "x in dict" means |
| 2022 | the same as dict.has_key(x). |
| 2023 | |
Barry Warsaw | 51acc8d | 2001-06-26 20:12:50 +0000 | [diff] [blame] | 2024 | - The update() method of dictionaries now accepts generic mapping |
| 2025 | objects. Specifically the argument object must support the .keys() |
| 2026 | and __getitem__() methods. This allows you to say, for example, |
| 2027 | {}.update(UserDict()) |
| 2028 | |
Guido van Rossum | 1031582 | 2001-05-01 20:54:30 +0000 | [diff] [blame] | 2029 | - Iterators were added; this is a generalized way of providing values |
| 2030 | to a for loop. See PEP 234. There's a new built-in function iter() |
| 2031 | to return an iterator. There's a new protocol to get the next value |
| 2032 | from an iterator using the next() method (in Python) or the |
| 2033 | tp_iternext slot (in C). There's a new protocol to get iterators |
| 2034 | using the __iter__() method (in Python) or the tp_iter slot (in C). |
| 2035 | Iterating (i.e. a for loop) over a dictionary generates its keys. |
| 2036 | Iterating over a file generates its lines. |
| 2037 | |
Tim Peters | f553f89 | 2001-05-01 20:45:31 +0000 | [diff] [blame] | 2038 | - The following functions were generalized to work nicely with iterator |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 2039 | arguments: :: |
| 2040 | |
Tim Peters | e63415e | 2001-05-08 04:38:29 +0000 | [diff] [blame] | 2041 | map(), filter(), reduce(), zip() |
Tim Peters | 8572b4f | 2001-05-06 01:05:02 +0000 | [diff] [blame] | 2042 | list(), tuple() (PySequence_Tuple() and PySequence_Fast() in C API) |
| 2043 | max(), min() |
Tim Peters | 1af03e9 | 2001-05-26 19:37:54 +0000 | [diff] [blame] | 2044 | join() method of strings |
| 2045 | extend() method of lists |
Tim Peters | 75f8e35 | 2001-05-05 11:33:43 +0000 | [diff] [blame] | 2046 | 'x in y' and 'x not in y' (PySequence_Contains() in C API) |
| 2047 | operator.countOf() (PySequence_Count() in C API) |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 2048 | right-hand side of assignment statements with multiple targets, such as :: |
Tim Peters | d6d010b | 2001-06-21 02:49:55 +0000 | [diff] [blame] | 2049 | x, y, z = some_iterable_object_returning_exactly_3_values |
Tim Peters | 75f8e35 | 2001-05-05 11:33:43 +0000 | [diff] [blame] | 2050 | |
Tim Peters | d85e102 | 2001-05-11 21:51:48 +0000 | [diff] [blame] | 2051 | - Accessing module attributes is significantly faster (for example, |
| 2052 | random.random or os.path or yourPythonModule.yourAttribute). |
| 2053 | |
Tim Peters | e63415e | 2001-05-08 04:38:29 +0000 | [diff] [blame] | 2054 | - Comparing dictionary objects via == and != is faster, and now works even |
| 2055 | if the keys and values don't support comparisons other than ==. |
| 2056 | |
Tim Peters | 95bf939 | 2001-05-10 08:32:44 +0000 | [diff] [blame] | 2057 | - Comparing dictionaries in ways other than == and != is slower: there were |
| 2058 | insecurities in the dict comparison implementation that could cause Python |
| 2059 | to crash if the element comparison routines for the dict keys and/or |
| 2060 | values mutated the dicts. Making the code bulletproof slowed it down. |
| 2061 | |
Tim Peters | eb28ef2 | 2001-06-02 05:27:19 +0000 | [diff] [blame] | 2062 | - Collisions in dicts are resolved via a new approach, which can help |
| 2063 | dramatically in bad cases. For example, looking up every key in a dict |
Tim Peters | 7a3bfc3 | 2001-06-12 01:22:22 +0000 | [diff] [blame] | 2064 | d with d.keys() == [i << 16 for i in range(20000)] is approximately 500x |
Tim Peters | eb28ef2 | 2001-06-02 05:27:19 +0000 | [diff] [blame] | 2065 | faster now. Thanks to Christian Tismer for pointing out the cause and |
| 2066 | the nature of an effective cure (last December! better late than never). |
Tim Peters | 15d4929 | 2001-05-27 07:39:22 +0000 | [diff] [blame] | 2067 | |
Tim Peters | 52e155e | 2001-06-16 05:42:57 +0000 | [diff] [blame] | 2068 | - repr() is much faster for large containers (dict, list, tuple). |
| 2069 | |
| 2070 | |
Tim Peters | a814db5 | 2001-05-14 07:05:58 +0000 | [diff] [blame] | 2071 | Library |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 2072 | ------- |
Tim Peters | a814db5 | 2001-05-14 07:05:58 +0000 | [diff] [blame] | 2073 | |
Fred Drake | 960fdf9 | 2001-07-20 18:38:26 +0000 | [diff] [blame] | 2074 | - The constants ascii_letters, ascii_lowercase. and ascii_uppercase |
Raymond Hettinger | f2f2a2c | 2002-08-19 14:25:03 +0000 | [diff] [blame] | 2075 | were added to the string module. These a locale-independent |
Fred Drake | 960fdf9 | 2001-07-20 18:38:26 +0000 | [diff] [blame] | 2076 | constants, unlike letters, lowercase, and uppercase. These are now |
| 2077 | use in appropriate locations in the standard library. |
| 2078 | |
Martin v. Löwis | f0473d5 | 2001-07-18 16:17:16 +0000 | [diff] [blame] | 2079 | - The flags used in dlopen calls can now be configured using |
| 2080 | sys.setdlopenflags and queried using sys.getdlopenflags. |
| 2081 | |
Guido van Rossum | c7e4aca | 2001-07-12 11:54:37 +0000 | [diff] [blame] | 2082 | - Fredrik Lundh's xmlrpclib is now a standard library module. This |
| 2083 | provides full client-side XML-RPC support. In addition, |
| 2084 | Demo/xmlrpc/ contains two server frameworks (one SocketServer-based, |
| 2085 | one asyncore-based). Thanks to Eric Raymond for the documentation. |
| 2086 | |
Guido van Rossum | 643d391 | 2001-07-05 14:46:25 +0000 | [diff] [blame] | 2087 | - The xrange() object is simplified: it no longer supports slicing, |
| 2088 | repetition, comparisons, efficient 'in' checking, the tolist() |
| 2089 | method, or the start, stop and step attributes. See PEP 260. |
| 2090 | |
Martin v. Löwis | ebf94db | 2001-06-06 06:25:40 +0000 | [diff] [blame] | 2091 | - A new function fnmatch.filter to filter lists of file names was added. |
| 2092 | |
Barry Warsaw | ffd674d | 2001-05-22 16:00:10 +0000 | [diff] [blame] | 2093 | - calendar.py uses month and day names based on the current locale. |
| 2094 | |
Guido van Rossum | 2e0a654 | 2001-05-15 02:14:44 +0000 | [diff] [blame] | 2095 | - strop is now *really* obsolete (this was announced before with 1.6), |
| 2096 | and issues DeprecationWarning when used (except for the four items |
| 2097 | that are still imported into string.py). |
| 2098 | |
Tim Peters | a814db5 | 2001-05-14 07:05:58 +0000 | [diff] [blame] | 2099 | - Cookie.py now sorts key+value pairs by key in output strings. |
| 2100 | |
| 2101 | - pprint.isrecursive(object) didn't correctly identify recursive objects. |
| 2102 | Now it does. |
| 2103 | |
Tim Peters | 95b3f78 | 2001-05-14 18:39:41 +0000 | [diff] [blame] | 2104 | - pprint functions now much faster for large containers (tuple, list, dict). |
| 2105 | |
Tim Peters | 7b9542a | 2001-06-10 23:40:19 +0000 | [diff] [blame] | 2106 | - New 'q' and 'Q' format codes in the struct module, corresponding to C |
| 2107 | types "long long" and "unsigned long long" (on Windows, __int64). In |
| 2108 | native mode, these can be used only when the platform C compiler supports |
| 2109 | these types (when HAVE_LONG_LONG is #define'd by the Python config |
| 2110 | process), and then they inherit the sizes and alignments of the C types. |
Tim Peters | 7a3bfc3 | 2001-06-12 01:22:22 +0000 | [diff] [blame] | 2111 | In standard mode, 'q' and 'Q' are supported on all platforms, and are |
| 2112 | 8-byte integral types. |
Tim Peters | 7b9542a | 2001-06-10 23:40:19 +0000 | [diff] [blame] | 2113 | |
Guido van Rossum | 83213cc | 2001-06-12 16:48:52 +0000 | [diff] [blame] | 2114 | - The site module installs a new built-in function 'help' that invokes |
| 2115 | pydoc.help. It must be invoked as 'help()'; when invoked as 'help', |
| 2116 | it displays a message reminding the user to use 'help()' or |
| 2117 | 'help(object)'. |
| 2118 | |
Tim Peters | a814db5 | 2001-05-14 07:05:58 +0000 | [diff] [blame] | 2119 | Tests |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 2120 | ----- |
Tim Peters | a814db5 | 2001-05-14 07:05:58 +0000 | [diff] [blame] | 2121 | |
| 2122 | - New test_mutants.py runs dict comparisons where the key and value |
Raymond Hettinger | f2f2a2c | 2002-08-19 14:25:03 +0000 | [diff] [blame] | 2123 | comparison operators mutate the dicts randomly during comparison. This |
Tim Peters | a814db5 | 2001-05-14 07:05:58 +0000 | [diff] [blame] | 2124 | rapidly causes Python to crash under earlier releases (not for the faint |
| 2125 | of heart: it can also cause Win9x to freeze or reboot!). |
| 2126 | |
Raymond Hettinger | f2f2a2c | 2002-08-19 14:25:03 +0000 | [diff] [blame] | 2127 | - New test_pprint.py verifies that pprint.isrecursive() and |
Tim Peters | 95b3f78 | 2001-05-14 18:39:41 +0000 | [diff] [blame] | 2128 | pprint.isreadable() return sensible results. Also verifies that simple |
| 2129 | cases produce correct output. |
Tim Peters | a814db5 | 2001-05-14 07:05:58 +0000 | [diff] [blame] | 2130 | |
Tim Peters | 4324aa3 | 2001-05-28 22:30:08 +0000 | [diff] [blame] | 2131 | C API |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 2132 | ----- |
Tim Peters | 4324aa3 | 2001-05-28 22:30:08 +0000 | [diff] [blame] | 2133 | |
| 2134 | - Removed the unused last_is_sticky argument from the internal |
| 2135 | _PyTuple_Resize(). If this affects you, you were cheating. |
Skip Montanaro | e5d7f7f | 2002-09-20 14:16:59 +0000 | [diff] [blame^] | 2136 | |
| 2137 | ---- |
| 2138 | |
| 2139 | **(For information about older versions, consult the HISTORY file.)** |