Python 3.0¶

The development cycle for Python versions 2.6 and 3.0 wassynchronized, with the alpha and beta releases for both versions beingmade on the same days. The development of 3.0 has influenced manyfeatures in 2.6.

Python 3.0 is a far-ranging redesign of Python that breakscompatibility with the 2.x series. This means that existing Pythoncode will need some conversion in order to run onPython 3.0. However, not all the changes in 3.0 necessarily breakcompatibility. In cases where new features won't cause existing codeto break, they've been backported to 2.6 and are described in thisdocument in the appropriate place. Some of the 3.0-derived featuresare:

• A__complex__() method for converting objects to a complex number.

• Alternate syntax for catching exceptions:exceptTypeErrorasexc.

• The addition offunctools.reduce() as a synonym for the built-inreduce() function.

Python 3.0 adds several new built-in functions and changes thesemantics of some existing builtins. Functions that are new in 3.0such asbin() have simply been added to Python 2.6, but existingbuiltins haven't been changed; instead, thefuture_builtinsmodule has versions with the new 3.0 semantics. Code written to becompatible with 3.0 can dofromfuture_builtinsimporthex,map asnecessary.

A new command-line switch,-3, enables warningsabout features that will be removed in Python 3.0. You can run codewith this switch to see how much work will be necessary to portcode to 3.0. The value of this switch is availableto Python code as the boolean variablesys.py3kwarning,and to C extension code asPy_Py3kWarningFlag.

Changes to the Development Process¶

While 2.6 was being developed, the Python development processunderwent two significant changes: we switched from SourceForge'sissue tracker to a customized Roundup installation, and thedocumentation was converted from LaTeX to reStructuredText.

PEP 343：'with' 陳述式¶

The previous version, Python 2.5, added the 'with'statement as an optional feature, to be enabled by afrom__future__importwith_statement directive. In 2.6 the statement no longer needs tobe specially enabled; this means thatwith is now always akeyword. The rest of this section is a copy of the correspondingsection from the "What's New in Python 2.5" document; if you'refamiliar with the 'with' statementfrom Python 2.5, you can skip this section.

The 'with' statement clarifies code that previously would usetry...finally blocks to ensure that clean-up code is executed. In thissection, I'll discuss the statement as it will commonly be used. In the nextsection, I'll examine the implementation details and show how to write objectsfor use with this statement.

The 'with' statement is a control-flow structure whose basicstructure is:

withexpression[asvariable]:with-block

The expression is evaluated, and it should result in an object that supports thecontext management protocol (that is, has__enter__() and__exit__()methods).

The object's__enter__() is called beforewith-block is executed andtherefore can run set-up code. It also may return a value that is bound to thenamevariable, if given. (Note carefully thatvariable isnot assignedthe result ofexpression.)

After execution of thewith-block is finished, the object's__exit__()method is called, even if the block raised an exception, and can therefore runclean-up code.

Some standard Python objects now support the context management protocol and canbe used with the 'with' statement. File objects are one example:

withopen('/etc/passwd','r')asf:forlineinf:printline...更多處理程式碼...

After this statement has executed, the file object inf will have beenautomatically closed, even if thefor loop raised an exceptionpart-way through the block.

Thethreading module's locks and condition variables also support the'with' statement:

lock=threading.Lock()withlock:# Critical section of code...

The lock is acquired before the block is executed and always released once theblock is complete.

Thelocalcontext() function in thedecimal module makes it easyto save and restore the current decimal context, which encapsulates the desiredprecision and rounding characteristics for computations:

fromdecimalimportDecimal,Context,localcontext# Displays with default precision of 28 digitsv=Decimal('578')printv.sqrt()withlocalcontext(Context(prec=16)):# All code in this block uses a precision of 16 digits.# The original context is restored on exiting the block.printv.sqrt()

db_connection=DatabaseConnection()withdb_connectionascursor:cursor.execute('insert into ...')cursor.execute('delete from ...')# ... 更多操作 ...

classDatabaseConnection:# Database interfacedefcursor(self):"Returns a cursor object and starts a new transaction"defcommit(self):"Commits current transaction"defrollback(self):"Rolls back current transaction"

classDatabaseConnection:...def__enter__(self):# Code to start a new transactioncursor=self.cursor()returncursor

classDatabaseConnection:...def__exit__(self,type,value,tb):iftbisNone:# No exception, so commitself.commit()else:# Exception occurred, so rollback.self.rollback()# return False

fromcontextlibimportcontextmanager@contextmanagerdefdb_transaction(connection):cursor=connection.cursor()try:yieldcursorexcept:connection.rollback()raiseelse:connection.commit()db=DatabaseConnection()withdb_transaction(db)ascursor:...

lock=threading.Lock()withnested(db_transaction(db),lock)as(cursor,locked):...

importurllib,sysfromcontextlibimportclosingwithclosing(urllib.urlopen('http://www.yahoo.com'))asf:forlineinf:sys.stdout.write(line)

PEP 366: Explicit Relative Imports From a Main Module¶

Python's-m switch allows running a module as a script.When you ran a module that was located inside a package, relativeimports didn't work correctly.

The fix for Python 2.6 adds amodule.__package__ attribute.When this attribute is present, relative imports will berelative to the value of this attribute instead of the__name__ attribute.

PEP 302-style importers can then set__package__ as necessary.Therunpy module that implements the-m switch nowdoes this, so relative imports will now work correctly in scriptsrunning from inside a package.

PEP 370: Per-usersite-packages Directory¶

When you run Python, the module search pathsys.path usuallyincludes a directory whose path ends in"site-packages". Thisdirectory is intended to hold locally installed packages available toall users using a machine or a particular site installation.

Python 2.6 introduces a convention for user-specific site directories.The directory varies depending on the platform:

• Unix 和 Mac OS X：~/.local/

• Windows：%APPDATA%/Python

Within this directory, there will be version-specific subdirectories,such aslib/python2.6/site-packages on Unix/Mac OS andPython26/site-packages on Windows.

If you don't like the default directory, it can be overridden by anenvironment variable.PYTHONUSERBASE sets the rootdirectory used for all Python versions supporting this feature. OnWindows, the directory for application-specific data can be changed bysetting theAPPDATA environment variable. You can alsomodify thesite.py file for your Python installation.

The feature can be disabled entirely by running Python with the-s option or setting thePYTHONNOUSERSITEenvironment variable.

PEP 371: Themultiprocessing Package¶

The newmultiprocessing package lets Python programs create newprocesses that will perform a computation and return a result to theparent. The parent and child processes can communicate using queuesand pipes, synchronize their operations using locks and semaphores,and can share simple arrays of data.

Themultiprocessing module started out as an exact emulation ofthethreading module using processes instead of threads. Thatgoal was discarded along the path to Python 2.6, but the generalapproach of the module is still similar. The fundamental classis theProcess, which is passed a callable object anda collection of arguments. Thestart() methodsets the callable running in a subprocess, after which you can calltheis_alive() method to check whether the subprocess is still runningand thejoin() method to wait for the process to exit.

Here's a simple example where the subprocess will calculate afactorial. The function doing the calculation is written strangely sothat it takes significantly longer when the input argument is amultiple of 4.

importtimefrommultiprocessingimportProcess,Queuedeffactorial(queue,N):"計算階乘。"# 如果 N 是 4 的倍數，則此函式將花費更多時間。if(N%4)==0:time.sleep(.05*N/4)# 計算結果fact=1Lforiinrange(1,N+1):fact=fact*i# 將結果放進佇列裡queue.put(fact)if__name__=='__main__':queue=Queue()N=5p=Process(target=factorial,args=(queue,N))p.start()p.join()result=queue.get()print'Factorial',N,'=',result

AQueue is used to communicate the result of the factorial.TheQueue object is stored in a global variable.The child process will use the value of the variable when the childwas created; because it's aQueue, parent and child can usethe object to communicate. (If the parent were to change the value ofthe global variable, the child's value would be unaffected, and viceversa.)

Two other classes,Pool andManager, providehigher-level interfaces.Pool will create a fixed number ofworker processes, and requests can then be distributed to the workersby callingapply() orapply_async() to add a single request,andmap() ormap_async() to add a number ofrequests. The following code uses aPool to spread requestsacross 5 worker processes and retrieve a list of results:

frommultiprocessingimportPooldeffactorial(N,dictionary):"計算階乘。"...p=Pool(5)result=p.map(factorial,range(1,1000,10))forvinresult:printv

This produces the following output:

13991680051090942171709440000822283865417792281772556288000000033452526613163807108170062053440751665152000000000...

The other high-level interface, theManager class, creates aseparate server process that can hold master copies of Python datastructures. Other processes can then access and modify these datastructures using proxy objects. The following example creates ashared dictionary by calling thedict() method; the workerprocesses then insert values into the dictionary. (Locking is notdone for you automatically, which doesn't matter in this example.Manager's methods also includeLock(),RLock(),andSemaphore() to create shared locks.)

importtimefrommultiprocessingimportPool,Managerdeffactorial(N,dictionary):"Compute a factorial."# Calculate the resultfact=1Lforiinrange(1,N+1):fact=fact*i# Store result in dictionarydictionary[N]=factif__name__=='__main__':p=Pool(5)mgr=Manager()d=mgr.dict()# Create shared dictionary# Run tasks using the poolforNinrange(1,1000,10):p.apply_async(factorial,(N,d))# Mark pool as closed -- no more tasks can be added.p.close()# Wait for tasks to exitp.join()# Output resultsfork,vinsorted(d.items()):printk,v

This will produce the output:

111139916800215109094217170944000031822283865417792281772556288000000041334525266131638071081700620534407516651520000000005115511187532873822802242430164693032110632597200169861120000...

PEP 3101: Advanced String Formatting¶

In Python 3.0, the% operator is supplemented by a more powerful stringformatting method,format(). Support for thestr.format() methodhas been backported to Python 2.6.

In 2.6, both 8-bit and Unicode strings have a.format() method thattreats the string as a template and takes the arguments to be formatted.The formatting template uses curly brackets ({,}) as special characters:

>>># Substitute positional argument 0 into the string.>>>"User ID:{0}".format("root")'User ID: root'>>># Use the named keyword arguments>>>"User ID:{uid}   Last seen:{last_login}".format(...uid="root",...last_login="5 Mar 2008 07:20")'User ID: root   Last seen: 5 Mar 2008 07:20'

Curly brackets can be escaped by doubling them:

>>>"Empty dict: {{}}".format()"Empty dict: {}"

Field names can be integers indicating positional arguments, such as{0},{1}, etc. or names of keyword arguments. You can alsosupply compound field names that read attributes or access dictionary keys:

>>>importsys>>>print'Platform:{0.platform}\nPython version:{0.version}'.format(sys)Platform: darwinPython version: 2.6a1+ (trunk:61261M, Mar  5 2008, 20:29:41)[GCC 4.0.1 (Apple Computer, Inc. build 5367)]'>>>importmimetypes>>>'Content-type:{0[.mp4]}'.format(mimetypes.types_map)'Content-type: video/mp4'

Note that when using dictionary-style notation such as[.mp4], youdon't need to put any quotation marks around the string; it will lookup the value using.mp4 as the key. Strings beginning with anumber will be converted to an integer. You can't write morecomplicated expressions inside a format string.

So far we've shown how to specify which field to substitute into theresulting string. The precise formatting used is also controllable byadding a colon followed by a format specifier. For example:

>>># Field 0: left justify, pad to 15 characters>>># Field 1: right justify, pad to 6 characters>>>fmt='{0:15} ${1:>6}'>>>fmt.format('Registration',35)'Registration    $    35'>>>fmt.format('Tutorial',50)'Tutorial        $    50'>>>fmt.format('Banquet',125)'Banquet         $   125'

Format specifiers can reference other fields through nesting:

>>>fmt='{0:{1}}'>>>width=15>>>fmt.format('Invoice #1234',width)'Invoice #1234  '>>>width=35>>>fmt.format('Invoice #1234',width)'Invoice #1234                      '

The alignment of a field within the desired width can be specified:

Format specifiers can also include a presentation type, whichcontrols how the value is formatted. For example, floating-point numberscan be formatted as a general number or in exponential notation:

>>>'{0:g}'.format(3.75)'3.75'>>>'{0:e}'.format(3.75)'3.750000e+00'

A variety of presentation types are available. Consult the 2.6documentation for acomplete list; here's a sample:

Classes and types can define a__format__() method to control how they'reformatted. It receives a single argument, the format specifier:

def__format__(self,format_spec):ifisinstance(format_spec,unicode):returnunicode(str(self))else:returnstr(self)

There's also aformat() builtin that will format a singlevalue. It calls the type's__format__() method with theprovided specifier:

>>>format(75.6564,'.2f')'75.66'

PEP 3105:print As a Function¶

Theprint statement becomes theprint() function in Python 3.0.Makingprint() a function makes it possible to replace the functionby doingdefprint(...) or importing a new function from somewhere else.

Python 2.6 has a__future__ import that removesprint as languagesyntax, letting you use the functional form instead. For example:

>>>from__future__importprint_function>>>print('# of entries',len(dictionary),file=sys.stderr)

The signature of the new function is:

defprint(*args,sep=' ',end='\n',file=None)

The parameters are:

• args: positional arguments whose values will be printed out.

• sep: the separator, which will be printed between arguments.

• end: the ending text, which will be printed after all of thearguments have been output.

• file: the file object to which the output will be sent.

PEP 3110: Exception-Handling Changes¶

One error that Python programmers occasionally makeis writing the following code:

try:...exceptTypeError,ValueError:# 錯誤！...

The author is probably trying to catch bothTypeError andValueError exceptions, but this code actually does somethingdifferent: it will catchTypeError and bind the resultingexception object to the local name"ValueError". TheValueError exception will not be caught at all. The correctcode specifies a tuple of exceptions:

try:...except(TypeError,ValueError):...

This error happens because the use of the comma here is ambiguous:does it indicate two different nodes in the parse tree, or a singlenode that's a tuple?

Python 3.0 makes this unambiguous by replacing the comma with the word"as". To catch an exception and store the exception object in thevariableexc, you must write:

try:...exceptTypeErrorasexc:...

Python 3.0 will only support the use of "as", and therefore interpretsthe first example as catching two different exceptions. Python 2.6supports both the comma and "as", so existing code will continue towork. We therefore suggest using "as" when writing new Python codethat will only be executed with 2.6.

PEP 3112: Byte Literals¶

Python 3.0 adopts Unicode as the language's fundamental string type anddenotes 8-bit literals differently, either asb'string'or using abytes constructor. For future compatibility,Python 2.6 addsbytes as a synonym for thestr type,and it also supports theb'' notation.

The 2.6str differs from 3.0'sbytes type in variousways; most notably, the constructor is completely different. In 3.0,bytes([65,66,67]) is 3 elements long, containing the bytesrepresentingABC; in 2.6,bytes([65,66,67]) returns the12-byte string representing thestr() of the list.

The primary use ofbytes in 2.6 will be to write tests ofobject type such asisinstance(x,bytes). This will help the 2to3converter, which can't tell whether 2.x code intends strings tocontain either characters or 8-bit bytes; you can nowuse eitherbytes orstr to represent your intentionexactly, and the resulting code will also be correct in Python 3.0.

There's also a__future__ import that causes all string literalsto become Unicode strings. This means that\u escape sequencescan be used to include Unicode characters:

from__future__importunicode_literalss=('\u751f\u3080\u304e\u3000\u751f\u3054''\u3081\u3000\u751f\u305f\u307e\u3054')printlen(s)# 12 個 Unicode 字元

At the C level, Python 3.0 will rename the existing 8-bitstring type, calledPyStringObject in Python 2.x,toPyBytesObject. Python 2.6 uses#defineto support using the namesPyBytesObject(),PyBytes_Check(),PyBytes_FromStringAndSize(),and all the other functions and macros used with strings.

Instances of thebytes type are immutable justas strings are. A newbytearray type stores a mutablesequence of bytes:

>>>bytearray([65,66,67])bytearray(b'ABC')>>>b=bytearray(u'\u21ef\u3244','utf-8')>>>bbytearray(b'\xe2\x87\xaf\xe3\x89\x84')>>>b[0]='\xe3'>>>bbytearray(b'\xe3\x87\xaf\xe3\x89\x84')>>>unicode(str(b),'utf-8')u'\u31ef \u3244'

Byte arrays support most of the methods of string types, such asstartswith()/endswith(),find()/rfind(),and some of the methods of lists, such asappend(),pop(), andreverse().

>>>b=bytearray('ABC')>>>b.append('d')>>>b.append(ord('e'))>>>bbytearray(b'ABCde')

There's also a corresponding C API, withPyByteArray_FromObject(),PyByteArray_FromStringAndSize(),and various other functions.

PEP 3116：新 I/O 函式庫¶

Python's built-in file objects support a number of methods, butfile-like objects don't necessarily support all of them. Objects thatimitate files usually supportread() andwrite(), but theymay not supportreadline(), for example. Python 3.0 introducesa layered I/O library in theio module that separates bufferingand text-handling features from the fundamental read and writeoperations.

There are three levels of abstract base classes provided bytheio module:

• RawIOBase defines raw I/O operations:read(),readinto(),write(),seek(),tell(),truncate(),andclose().Most of the methods of this class will often map to a single system call.There are alsoreadable(),writable(), andseekable()methods for determining what operations a given object will allow.

  Python 3.0 has concrete implementations of this class for files andsockets, but Python 2.6 hasn't restructured its file and socket objectsin this way.

• BufferedIOBase is an abstract base class thatbuffers data in memory to reduce the number ofsystem calls used, making I/O processing more efficient.It supports all of the methods ofRawIOBase,and adds araw attribute holding the underlying raw object.

  There are five concrete classes implementing this ABC.BufferedWriter andBufferedReader are for objectsthat support write-only or read-only usage that have aseek()method for random access.BufferedRandom objects supportread and write access upon the same underlying stream, andBufferedRWPair is for objects such as TTYs that have bothread and write operations acting upon unconnected streams of data.TheBytesIO class supports reading, writing, and seekingover an in-memory buffer.

• TextIOBase: Provides functions for reading and writingstrings (remember, strings will be Unicode in Python 3.0),and supportinguniversal newlines.TextIOBase definesthereadline() method and supports iteration uponobjects.

  There are two concrete implementations.TextIOWrapperwraps a buffered I/O object, supporting all of the methods fortext I/O and adding abuffer attribute for accessto the underlying object.StringIO simply bufferseverything in memory without ever writing anything to disk.

  (In Python 2.6,io.StringIO is implemented inpure Python, so it's pretty slow. You should therefore stick with theexistingStringIO module orcStringIO for now. At somepoint Python 3.0'sio module will be rewritten into C for speed,and perhaps the C implementation will be backported to the 2.x releases.)

In Python 2.6, the underlying implementations haven't beenrestructured to build on top of theio module's classes. Themodule is being provided to make it easier to write code that'sforward-compatible with 3.0, and to save developers the effort of writingtheir own implementations of buffering and text I/O.

PEP 3118：修訂緩衝協定¶

The buffer protocol is a C-level API that lets Python typesexchange pointers into their internal representations. Amemory-mapped file can be viewed as a buffer of characters, forexample, and this lets another module such asretreat memory-mapped files as a string of characters to be searched.

The primary users of the buffer protocol are numeric-processingpackages such as NumPy, which expose the internal representationof arrays so that callers can write data directly into an array insteadof going through a slower API. This PEP updates the buffer protocol in light of experiencefrom NumPy development, adding a number of new featuressuch as indicating the shape of an array or locking a memory region.

The most important new C API function isPyObject_GetBuffer(PyObject*obj,Py_buffer*view,intflags), whichtakes an object and a set of flags, and fills in thePy_buffer structure with informationabout the object's memory representation. Objectscan use this operation to lock memory in placewhile an external caller could be modifying the contents,so there's a correspondingPyBuffer_Release(Py_buffer*view) toindicate that the external caller is done.

Theflags argument toPyObject_GetBuffer() specifiesconstraints upon the memory returned. Some examples are:

• PyBUF_WRITABLE indicates that the memory must be writable.

• PyBUF_LOCK requests a read-only or exclusive lock on the memory.

• PyBUF_C_CONTIGUOUS andPyBUF_F_CONTIGUOUSrequests a C-contiguous (last dimension varies the fastest) orFortran-contiguous (first dimension varies the fastest) array layout.

Two new argument codes forPyArg_ParseTuple(),s* andz*, return locked buffer objects for a parameter.

PEP 3119: Abstract Base Classes¶

Some object-oriented languages such as Java support interfaces,declaring that a class has a given set of methods or supports a givenaccess protocol. Abstract Base Classes (or ABCs) are an equivalentfeature for Python. The ABC support consists of anabc modulecontaining a metaclass calledABCMeta, special handling ofthis metaclass by theisinstance() andissubclass()builtins, and a collection of basic ABCs that the Python developersthink will be widely useful. Future versions of Python will probablyadd more ABCs.

Let's say you have a particular class and wish to know whether it supportsdictionary-style access. The phrase "dictionary-style" is vague, however.It probably means that accessing items withobj[1] works.Does it imply that setting items withobj[2]=value works?Or that the object will havekeys(),values(), anditems()methods? What about the iterative variants such asiterkeys()?copy()andupdate()? Iterating over the object withiter()?

The Python 2.6collections module includes a number ofdifferent ABCs that represent these distinctions.Iterableindicates that a class defines__iter__(), andContainer means the class defines a__contains__()method and therefore supportsxiny expressions. The basicdictionary interface of getting items, setting items, andkeys(),values(), anditems(), is defined by theMutableMapping ABC.

You can derive your own classes from a particular ABCto indicate they support that ABC's interface:

importcollectionsclassStorage(collections.MutableMapping):...

Alternatively, you could write the class without deriving fromthe desired ABC and instead register the class bycalling the ABC'sregister() method:

importcollectionsclassStorage:...collections.MutableMapping.register(Storage)

For classes that you write, deriving from the ABC is probably clearer.Theregister() method is useful when you've written a newABC that can describe an existing type or class, or if you wantto declare that some third-party class implements an ABC.For example, if you defined aPrintableType ABC,it's legal to do:

# 註冊 Python 的型別PrintableType.register(int)PrintableType.register(float)PrintableType.register(str)

Classes should obey the semantics specified by an ABC, butPython can't check this; it's up to the class author tounderstand the ABC's requirements and to implement the code accordingly.

To check whether an object supports a particular interface, you cannow write:

deffunc(d):ifnotisinstance(d,collections.MutableMapping):raiseValueError("Mapping object expected, not%r"%d)

Don't feel that you must now begin writing lots of checks as in theabove example. Python has a strong tradition of duck-typing, whereexplicit type-checking is never done and code simply calls methods onan object, trusting that those methods will be there and raising anexception if they aren't. Be judicious in checking for ABCs and onlydo it where it's absolutely necessary.

You can write your own ABCs by usingabc.ABCMeta as themetaclass in a class definition:

fromabcimportABCMeta,abstractmethodclassDrawable():__metaclass__=ABCMeta@abstractmethoddefdraw(self,x,y,scale=1.0):passdefdraw_doubled(self,x,y):self.draw(x,y,scale=2.0)classSquare(Drawable):defdraw(self,x,y,scale):...

In theDrawable ABC above, thedraw_doubled() methodrenders the object at twice its size and can be implemented in termsof other methods described inDrawable. Classes implementingthis ABC therefore don't need to provide their own implementationofdraw_doubled(), though they can do so. An implementationofdraw() is necessary, though; the ABC can't providea useful generic implementation.

You can apply the@abstractmethod decorator to methods such asdraw() that must be implemented; Python will then raise anexception for classes that don't define the method.Note that the exception is only raised when you actuallytry to create an instance of a subclass lacking the method:

>>>classCircle(Drawable):...pass...>>>c=Circle()Traceback (most recent call last):  File"<stdin>", line1, in<module>TypeError:Can't instantiate abstract class Circle with abstract methods draw>>>

Abstract data attributes can be declared using the@abstractproperty decorator:

fromabcimportabstractproperty...@abstractpropertydefreadonly(self):returnself._x

Subclasses must then define areadonly() property.

PEP 3127: Integer Literal Support and Syntax¶

Python 3.0 changes the syntax for octal (base-8) integer literals,prefixing them with "0o" or "0O" instead of a leading zero, and addssupport for binary (base-2) integer literals, signalled by a "0b" or"0B" prefix.

Python 2.6 doesn't drop support for a leading 0 signallingan octal number, but it does add support for "0o" and "0b":

>>>0o21,2*8+1(17, 17)>>>0b10111147

Theoct() builtin still returns numbersprefixed with a leading zero, and a newbin()builtin returns the binary representation for a number:

>>>oct(42)'052'>>>future_builtins.oct(42)'0o52'>>>bin(173)'0b10101101'

Theint() andlong() builtins will now accept the "0o"and "0b" prefixes when base-8 or base-2 are requested, or when thebase argument is zero (signalling that the base used should bedetermined from the string):

>>>int('0o52',0)42>>>int('1101',2)13>>>int('0b1101',2)13>>>int('0b1101',0)13

PEP 3129: Class Decorators¶

Decorators have been extended from functions to classes. It's now legal towrite:

@foo@barclassA:pass

這等價於：

classA:passA=foo(bar(A))

PEP 3141: A Type Hierarchy for Numbers¶

Python 3.0 adds several abstract base classes for numeric typesinspired by Scheme's numeric tower. These classes were backported to2.6 as thenumbers module.

The most general ABC isNumber. It defines no operations atall, and only exists to allow checking if an object is a number bydoingisinstance(obj,Number).

Complex is a subclass ofNumber. Complex numberscan undergo the basic operations of addition, subtraction,multiplication, division, and exponentiation, and you can retrieve thereal and imaginary parts and obtain a number's conjugate. Python's built-incomplex type is an implementation ofComplex.

Real further derives fromComplex, and addsoperations that only work on real numbers:floor(),trunc(),rounding, taking the remainder mod N, floor division,and comparisons.

Rational numbers derive fromReal, havenumerator anddenominator properties, and can beconverted to floats. Python 2.6 adds a simple rational-number class,Fraction, in thefractions module. (It's calledFraction instead ofRational to avoida name clash withnumbers.Rational.)

Integral numbers derive fromRational, andcan be shifted left and right with<< and>>,combined using bitwise operations such as& and|,and can be used as array indexes and slice boundaries.

In Python 3.0, the PEP slightly redefines the existing builtinsround(),math.floor(),math.ceil(), and adds a newone,math.trunc(), that's been backported to Python 2.6.math.trunc() rounds toward zero, returning the closestIntegral that's between the function's argument and zero.

>>>fromfractionsimportFraction>>>a=Fraction(2,3)>>>b=Fraction(2,5)>>>float(a),float(b)(0.66666666666666663, 0.40000000000000002)>>>a+bFraction(16, 15)>>>a/bFraction(5, 3)

>>>(2.5).as_integer_ratio()(5, 2)>>>(3.1415).as_integer_ratio()(7074029114692207L, 2251799813685248L)>>>(1./3).as_integer_ratio()(6004799503160661L, 18014398509481984L)

其他語言更動¶

Some smaller changes made to the core Python language are:

• Directories and zip archives containing a__main__.py filecan now be executed directly by passing their name to theinterpreter. The directory or zip archive is automatically insertedas the first entry in sys.path. (Suggestion and initial patch byAndy Chu, subsequently revised by Phillip J. Eby and Nick Coghlan;bpo-1739468.)

• Thehasattr() function was catching and ignoring all errors,under the assumption that they meant a__getattr__() methodwas failing somehow and the return value ofhasattr() wouldtherefore beFalse. This logic shouldn't be applied toKeyboardInterrupt andSystemExit, however; Python 2.6will no longer discard such exceptions whenhasattr()encounters them. (Fixed by Benjamin Peterson;bpo-2196.)

• When calling a function using the** syntax to provide keywordarguments, you are no longer required to use a Python dictionary;any mapping will now work:

  >>>deff(**kw):...printsorted(kw)...>>>ud=UserDict.UserDict()>>>ud['a']=1>>>ud['b']='string'>>>f(**ud)['a', 'b']

  （由 Alexander Belopolsky 所貢獻；bpo-1686487。）

  It's also become legal to provide keyword arguments after a*args argumentto a function call.

  >>>deff(*args,**kw):...printargs,kw...>>>f(1,2,3,*(4,5,6),keyword=13)(1, 2, 3, 4, 5, 6) {'keyword': 13}

  Previously this would have been a syntax error.(Contributed by Amaury Forgeot d'Arc;bpo-3473.)

• A new builtin,next(iterator,[default]) returns the next itemfrom the specified iterator. If thedefault argument is supplied,it will be returned ifiterator has been exhausted; otherwise,theStopIteration exception will be raised. (Backportedinbpo-2719.)

• Tuples now haveindex() andcount() methods matching thelist type'sindex() andcount() methods:

  >>>t=(0,1,2,3,4,0,1,2)>>>t.index(3)3>>>t.count(0)2

  （由 Raymond Hettinger 所貢獻。）

• The built-in types now have improved support for extended slicing syntax,accepting various combinations of(start,stop,step).Previously, the support was partial and certain corner cases wouldn't work.(Implemented by Thomas Wouters.)

• Properties now have three attributes,getter,setteranddeleter, that are decorators providing useful shortcutsfor adding a getter, setter or deleter function to an existingproperty. You would use them like this:

  classC(object):@propertydefx(self):returnself._x@x.setterdefx(self,value):self._x=value@x.deleterdefx(self):delself._xclassD(C):@C.x.getterdefx(self):returnself._x*2@x.setterdefx(self,value):self._x=value/2

• Several methods of the built-in set types now accept multiple iterables:intersection(),intersection_update(),union(),update(),difference() anddifference_update().

  >>>s=set('1234567890')>>>s.intersection('abc123','cdf246')# Intersection between all inputsset(['2'])>>>s.difference('246','789')set(['1', '0', '3', '5'])

  （由 Raymond Hettinger 所貢獻。）

• Many floating-point features were added. Thefloat() functionwill now turn the stringnan into anIEEE 754 Not A Number value, and+inf and-inf intopositive or negative infinity. This works on any platform withIEEE 754 semantics. (Contributed by Christian Heimes;bpo-1635.)

  Other functions in themath module,isinf() andisnan(), return true if their floating-point argument isinfinite or Not A Number. (bpo-1640)

  Conversion functions were added to convert floating-point numbersinto hexadecimal strings (bpo-3008). These functionsconvert floats to and from a string representation withoutintroducing rounding errors from the conversion between decimal andbinary. Floats have ahex() method that returns a stringrepresentation, and thefloat.fromhex() method converts a stringback into a number:

  >>>a=3.75>>>a.hex()'0x1.e000000000000p+1'>>>float.fromhex('0x1.e000000000000p+1')3.75>>>b=1./3>>>b.hex()'0x1.5555555555555p-2'

• A numerical nicety: when creating a complex number from two floatson systems that support signed zeros (-0 and +0), thecomplex() constructor will now preserve the signof the zero. (Fixed by Mark T. Dickinson;bpo-1507.)

• Classes that inherit a__hash__() method from a parent classcan set__hash__=None to indicate that the class isn'thashable. This will makehash(obj) raise aTypeErrorand the class will not be indicated as implementing theHashable ABC.

  You should do this when you've defined a__cmp__() or__eq__() method that compares objects by their value ratherthan by identity. All objects have a default hash method that usesid(obj) as the hash value. There's no tidy way to remove the__hash__() method inherited from a parent class, soassigningNone was implemented as an override. At theC level, extensions can settp_hash toPyObject_HashNotImplemented().(Fixed by Nick Coghlan and Amaury Forgeot d'Arc;bpo-2235.)

• TheGeneratorExit exception now subclassesBaseException instead ofException. This meansthat an exception handler that doesexceptException:will not inadvertently catchGeneratorExit.(Contributed by Chad Austin;bpo-1537.)

• Generator objects now have agi_code attribute that refers tothe original code object backing the generator.(Contributed by Collin Winter;bpo-1473257.)

• Thecompile() built-in function now accepts keyword argumentsas well as positional parameters. (Contributed by Thomas Wouters;bpo-1444529.)

• Thecomplex() constructor now accepts strings containingparenthesized complex numbers, meaning thatcomplex(repr(cplx))will now round-trip values. For example,complex('(3+4j)')now returns the value (3+4j). (bpo-1491866)

• The stringtranslate() method now acceptsNone as thetranslation table parameter, which is treated as the identitytransformation. This makes it easier to carry out operationsthat only delete characters. (Contributed by Bengt Richter andimplemented by Raymond Hettinger;bpo-1193128.)

• The built-indir() function now checks for a__dir__()method on the objects it receives. This method must return a listof strings containing the names of valid attributes for the object,and lets the object control the value thatdir() produces.Objects that have__getattr__() or__getattribute__()methods can use this to advertise pseudo-attributes they will honor.(bpo-1591665)

• Instance method objects have new attributes for the object and functioncomprising the method; the new synonym forim_self is__self__, andim_func is also available as__func__.The old names are still supported in Python 2.6, but are gone in 3.0.

• An obscure change: when you use thelocals() function inside aclass statement, the resulting dictionary no longer returns freevariables. (Free variables, in this case, are variables referenced in theclass statement that aren't attributes of the class.)

New and Improved Modules¶

As in every release, Python's standard library received a number ofenhancements and bug fixes. Here's a partial list of the most notablechanges, sorted alphabetically by module name. Consult theMisc/NEWS file in the source tree for a more complete list ofchanges, or look through the Subversion logs for all the details.

• Theasyncore andasynchat modules arebeing actively maintained again, and a number of patches and bugfixeswere applied. (Maintained by Josiah Carlson; seebpo-1736190 forone patch.)

• Thebsddb module also has a new maintainer, Jesús Cea Avión, and the packageis now available as a standalone package. The web page for the package iswww.jcea.es/programacion/pybsddb.htm.The plan is to remove the package from the standard libraryin Python 3.0, because its pace of releases is much more frequent thanPython's.

  Thebsddb.dbshelve module now uses the highest pickling protocolavailable, instead of restricting itself to protocol 1.(Contributed by W. Barnes.)

• Thecgi module will now read variables from the query stringof an HTTP POST request. This makes it possible to use form actionswith URLs that include query strings such as"/cgi-bin/add.py?category=1". (Contributed by Alexandre Fiori andNubis;bpo-1817.)

  Theparse_qs() andparse_qsl() functions have beenrelocated from thecgi module to theurlparse module.The versions still available in thecgi module willtriggerPendingDeprecationWarning messages in 2.6(bpo-600362).

• Thecmath module underwent extensive revision,contributed by Mark Dickinson and Christian Heimes.Five new functions were added:

  • polar() converts a complex number to polar form, returningthe modulus and argument of the complex number.

  • rect() does the opposite, turning a modulus, argument pairback into the corresponding complex number.

  • phase() returns the argument (also called the angle) of a complexnumber.

  • isnan() returns True if eitherthe real or imaginary part of its argument is a NaN.

  • isinf() returns True if either the real or imaginary part ofits argument is infinite.

  The revisions also improved the numerical soundness of thecmath module. For all functions, the real and imaginaryparts of the results are accurate to within a few units of leastprecision (ulps) whenever possible. Seebpo-1381 for thedetails. The branch cuts forasinh(),atanh(): andatan() have also been corrected.

  The tests for the module have been greatly expanded; nearly 2000 newtest cases exercise the algebraic functions.

  On IEEE 754 platforms, thecmath module now handles IEEE 754special values and floating-point exceptions in a manner consistentwith Annex 'G' of the C99 standard.

• A new data type in thecollections module:namedtuple(typename,fieldnames)is a factory function that creates subclasses of the standard tuplewhose fields are accessible by name as well as index. For example:

  >>>var_type=collections.namedtuple('variable',...'id name type size')>>># Names are separated by spaces or commas.>>># 'id, name, type, size' would also work.>>>var_type._fields('id', 'name', 'type', 'size')>>>var=var_type(1,'frequency','int',4)>>>printvar[0],var.id# Equivalent1 1>>>printvar[2],var.type# Equivalentint int>>>var._asdict(){'size': 4, 'type': 'int', 'id': 1, 'name': 'frequency'}>>>v2=var._replace(name='amplitude')>>>v2variable(id=1, name='amplitude', type='int', size=4)

  Several places in the standard library that returned tuples havebeen modified to returnnamedtuple() instances. For example,theDecimal.as_tuple() method now returns a named tuple withsign,digits, andexponent fields.

  （由 Raymond Hettinger 所貢獻。）

• Another change to thecollections module is that thedeque type now supports an optionalmaxlen parameter;if supplied, the deque's size will be restricted to no morethanmaxlen items. Adding more items to a full deque causesold items to be discarded.

  >>>fromcollectionsimportdeque>>>dq=deque(maxlen=3)>>>dqdeque([], maxlen=3)>>>dq.append(1);dq.append(2);dq.append(3)>>>dqdeque([1, 2, 3], maxlen=3)>>>dq.append(4)>>>dqdeque([2, 3, 4], maxlen=3)

  （由 Raymond Hettinger 所貢獻。）

• TheCookie module'sMorsel objects now support anhttponly attribute. In some browsers. cookies with this attributeset cannot be accessed or manipulated by JavaScript code.(Contributed by Arvin Schnell;bpo-1638033.)

• A new window method in thecurses module,chgat(), changes the display attributes for a certain number ofcharacters on a single line. (Contributed by Fabian Kreutz.)

  # Boldface text starting at y=0,x=21# and affecting the rest of the line.stdscr.chgat(0,21,curses.A_BOLD)

  TheTextbox class in thecurses.textpad modulenow supports editing in insert mode as well as overwrite mode.Insert mode is enabled by supplying a true value for theinsert_modeparameter when creating theTextbox instance.

• Thedatetime module'sstrftime() methods now support a%f format code that expands to the number of microseconds in theobject, zero-padded onthe left to six places. (Contributed by Skip Montanaro;bpo-1158.)

• Thedecimal module was updated to version 1.66 ofthe General Decimal Specification. New featuresinclude some methods for some basic mathematical functions such asexp() andlog10():

  >>>Decimal(1).exp()Decimal("2.718281828459045235360287471")>>>Decimal("2.7182818").ln()Decimal("0.9999999895305022877376682436")>>>Decimal(1000).log10()Decimal("3")

  Theas_tuple() method ofDecimal objects now returns anamed tuple withsign,digits, andexponent fields.

  (Implemented by Facundo Batista and Mark Dickinson. Named tuplesupport added by Raymond Hettinger.)

• Thedifflib module'sSequenceMatcher classnow returns named tuples representing matches,witha,b, andsize attributes.(Contributed by Raymond Hettinger.)

• An optionaltimeout parameter, specifying a timeout measured inseconds, was added to theftplib.FTP class constructor aswell as theconnect() method. (Added by Facundo Batista.)Also, theFTP class'sstorbinary() andstorlines() now take an optionalcallback parameter thatwill be called with each block of data after the data has been sent.(Contributed by Phil Schwartz;bpo-1221598.)

• Thereduce() built-in function is also available in thefunctools module. In Python 3.0, the builtin has beendropped andreduce() is only available fromfunctools;currently there are no plans to drop the builtin in the 2.x series.(Patched by Christian Heimes;bpo-1739906.)

• When possible, thegetpass module will now use/dev/tty to print a prompt message and read the password,falling back to standard error and standard input. If thepassword may be echoed to the terminal, a warning is printed beforethe prompt is displayed. (Contributed by Gregory P. Smith.)

• Theglob.glob() function can now return Unicode filenames ifa Unicode path was used and Unicode filenames are matched within thedirectory. (bpo-1001604)

• A new function in theheapq module,merge(iter1,iter2,...),takes any number of iterables returning data in sortedorder, and returns a new generator that returns the contents of allthe iterators, also in sorted order. For example:

  >>>list(heapq.merge([1,3,5,9],[2,8,16]))[1, 2, 3, 5, 8, 9, 16]

  Another new function,heappushpop(heap,item),pushesitem ontoheap, then pops off and returns the smallest item.This is more efficient than making a call toheappush() and thenheappop().

  heapq is now implemented to only use less-than comparison,instead of the less-than-or-equal comparison it previously used.This makesheapq's usage of a type match thelist.sort() method.(Contributed by Raymond Hettinger.)

• An optionaltimeout parameter, specifying a timeout measured inseconds, was added to thehttplib.HTTPConnection andHTTPSConnection class constructors. (Added by FacundoBatista.)

• Most of theinspect module's functions, such asgetmoduleinfo() andgetargs(), now return named tuples.In addition to behaving like tuples, the elements of the return valuecan also be accessed as attributes.(Contributed by Raymond Hettinger.)

  Some new functions in the module includeisgenerator(),isgeneratorfunction(),andisabstract().

• Theitertools module gained several new functions.

  izip_longest(iter1,iter2,...[,fillvalue]) makes tuples fromeach of the elements; if some of the iterables are shorter thanothers, the missing values are set tofillvalue. For example:

  >>>tuple(itertools.izip_longest([1,2,3],[1,2,3,4,5]))((1, 1), (2, 2), (3, 3), (None, 4), (None, 5))

  product(iter1,iter2,...,[repeat=N]) returns the Cartesian productof the supplied iterables, a set of tuples containingevery possible combination of the elements returned from each iterable.

  >>>list(itertools.product([1,2,3],[4,5,6]))[(1, 4), (1, 5), (1, 6), (2, 4), (2, 5), (2, 6), (3, 4), (3, 5), (3, 6)]

  The optionalrepeat keyword argument is used for taking theproduct of an iterable or a set of iterables with themselves,repeatedN times. With a single iterable argument,N-tuplesare returned:

  >>>list(itertools.product([1,2],repeat=3))[(1, 1, 1), (1, 1, 2), (1, 2, 1), (1, 2, 2), (2, 1, 1), (2, 1, 2), (2, 2, 1), (2, 2, 2)]

  With two iterables,2N-tuples are returned.

  >>>list(itertools.product([1,2],[3,4],repeat=2))[(1, 3, 1, 3), (1, 3, 1, 4), (1, 3, 2, 3), (1, 3, 2, 4), (1, 4, 1, 3), (1, 4, 1, 4), (1, 4, 2, 3), (1, 4, 2, 4), (2, 3, 1, 3), (2, 3, 1, 4), (2, 3, 2, 3), (2, 3, 2, 4), (2, 4, 1, 3), (2, 4, 1, 4), (2, 4, 2, 3), (2, 4, 2, 4)]

  combinations(iterable,r) returns sub-sequences of lengthr fromthe elements ofiterable.

  >>>list(itertools.combinations('123',2))[('1', '2'), ('1', '3'), ('2', '3')]>>>list(itertools.combinations('123',3))[('1', '2', '3')]>>>list(itertools.combinations('1234',3))[('1', '2', '3'), ('1', '2', '4'), ('1', '3', '4'), ('2', '3', '4')]

  permutations(iter[,r]) returns all the permutations of lengthr ofthe iterable's elements. Ifr is not specified, it will default to thenumber of elements produced by the iterable.

  >>>list(itertools.permutations([1,2,3,4],2))[(1, 2), (1, 3), (1, 4), (2, 1), (2, 3), (2, 4), (3, 1), (3, 2), (3, 4), (4, 1), (4, 2), (4, 3)]

  itertools.chain(*iterables) is an existing function initertools that gained a new constructor in Python 2.6.itertools.chain.from_iterable(iterable) takes a singleiterable that should return other iterables.chain() willthen return all the elements of the first iterable, thenall the elements of the second, and so on.

  >>>list(itertools.chain.from_iterable([[1,2,3],[4,5,6]]))[1, 2, 3, 4, 5, 6]

  (All contributed by Raymond Hettinger.)

• Thelogging module'sFileHandler classand its subclassesWatchedFileHandler,RotatingFileHandler,andTimedRotatingFileHandler nowhave an optionaldelay parameter to their constructors. Ifdelayis true, opening of the log file is deferred until the firstemit() call is made. (Contributed by Vinay Sajip.)

  TimedRotatingFileHandler also has autc constructorparameter. If the argument is true, UTC time will be usedin determining when midnight occurs and in generating filenames;otherwise local time will be used.

• Several new functions were added to themath module:

  • isinf() andisnan() determine whether a given floatis a (positive or negative) infinity or a NaN (Not a Number), respectively.

  • copysign() copies the sign bit of an IEEE 754 number,returning the absolute value ofx combined with the sign bit ofy. For example,math.copysign(1,-0.0) returns -1.0.(Contributed by Christian Heimes.)

  • factorial() computes the factorial of a number.(Contributed by Raymond Hettinger;bpo-2138.)

  • fsum() adds up the stream of numbers from an iterable,and is careful to avoid loss of precision through using partial sums.(Contributed by Jean Brouwers, Raymond Hettinger, and Mark Dickinson;bpo-2819.)

  • acosh(),asinh()andatanh() compute the inverse hyperbolic functions.

  • log1p() returns the natural logarithm of1+x(basee).

  • trunc() rounds a number toward zero, returning the closestIntegral that's between the function's argument and zero.Added as part of the backport ofPEP 3141's type hierarchy for numbers.

• Themath module has been improved to give more consistentbehaviour across platforms, especially with respect to handling offloating-point exceptions and IEEE 754 special values.

  Whenever possible, the module follows the recommendations of the C99standard about 754's special values. For example,sqrt(-1.)should now give aValueError across almost all platforms,whilesqrt(float('NaN')) should return a NaN on all IEEE 754platforms. Where Annex 'F' of the C99 standard recommends signaling'divide-by-zero' or 'invalid', Python will raiseValueError.Where Annex 'F' of the C99 standard recommends signaling 'overflow',Python will raiseOverflowError. (Seebpo-711019 andbpo-1640.)

  （由 Christian Heimes 和 Mark Dickinson 所貢獻。）

• mmap objects now have arfind() method that searches for asubstring beginning at the end of the string and searchingbackwards. Thefind() method also gained anend parametergiving an index at which to stop searching.(Contributed by John Lenton.)

• Theoperator module gained amethodcaller() function that takes a name and an optionalset of arguments, returning a callable that will callthe named function on any arguments passed to it. For example:

  >>># Equivalent to lambda s: s.replace('old', 'new')>>>replacer=operator.methodcaller('replace','old','new')>>>replacer('old wine in old bottles')'new wine in new bottles'

  （經 Gregory Petrosyan 建議後由 Georg Brandl 所貢獻。）

  Theattrgetter() function now accepts dotted names and performsthe corresponding attribute lookups:

  >>>inst_name=operator.attrgetter(...'__class__.__name__')>>>inst_name('')'str'>>>inst_name(help)'_Helper'

  （經 Barry Warsaw 建議後由 Georg Brandl 所貢獻。）

• Theos module now wraps several new system calls.fchmod(fd,mode) andfchown(fd,uid,gid) change the modeand ownership of an opened file, andlchmod(path,mode) changesthe mode of a symlink. (Contributed by Georg Brandl and ChristianHeimes.)

  chflags() andlchflags() are wrappers for thecorresponding system calls (where they're available), changing theflags set on a file. Constants for the flag values are defined inthestat module; some possible values includeUF_IMMUTABLE to signal the file may not be changed andUF_APPEND to indicate that data can only be appended to thefile. (Contributed by M. Levinson.)

  os.closerange(low,high) efficiently closes all file descriptorsfromlow tohigh, ignoring any errors and not includinghigh itself.This function is now used by thesubprocess module to make startingprocesses faster. (Contributed by Georg Brandl;bpo-1663329.)

• Theos.environ object'sclear() method will now unset theenvironment variables usingos.unsetenv() in addition to clearingthe object's keys. (Contributed by Martin Horcicka;bpo-1181.)

• Theos.walk() function now has afollowlinks parameter. Ifset to True, it will follow symlinks pointing to directories andvisit the directory's contents. For backward compatibility, theparameter's default value is false. Note that the function can fallinto an infinite recursion if there's a symlink that points to aparent directory. (bpo-1273829)

• In theos.path module, thesplitext() functionhas been changed to not split on leading period characters.This produces better results when operating on Unix's dot-files.For example,os.path.splitext('.ipython')now returns('.ipython','') instead of('','.ipython').(bpo-1115886)

  A new function,os.path.relpath(path,start='.'), returns a relative pathfrom thestart path, if it's supplied, or from the currentworking directory to the destinationpath. (Contributed byRichard Barran;bpo-1339796.)

  On Windows,os.path.expandvars() will now expand environment variablesgiven in the form "%var%", and "~user" will be expanded into theuser's home directory path. (Contributed by Josiah Carlson;bpo-957650.)

• The Python debugger provided by thepdb modulegained a new command: "run" restarts the Python program being debuggedand can optionally take new command-line arguments for the program.(Contributed by Rocky Bernstein;bpo-1393667.)

• Thepdb.post_mortem() function, used to begin debugging atraceback, will now use the traceback returned bysys.exc_info()if no traceback is supplied. (Contributed by Facundo Batista;bpo-1106316.)

• Thepickletools module now has anoptimize() functionthat takes a string containing a pickle and removes some unusedopcodes, returning a shorter pickle that contains the same data structure.(Contributed by Raymond Hettinger.)

• Aget_data() function was added to thepkgutilmodule that returns the contents of resource files includedwith an installed Python package. For example:

  >>>importpkgutil>>>printpkgutil.get_data('test','exception_hierarchy.txt')BaseException +-- SystemExit +-- KeyboardInterrupt +-- GeneratorExit +-- Exception      +-- StopIteration      +-- StandardError ...

  （由 Paul Moore 貢獻；bpo-2439。）

• Thepyexpat module'sParser objects now allow settingtheirbuffer_size attribute to change the size of the bufferused to hold character data.(Contributed by Achim Gaedke;bpo-1137.)

• TheQueue module now provides queue variants that retrieve entriesin different orders. ThePriorityQueue class storesqueued items in a heap and retrieves them in priority order,andLifoQueue retrieves the most recently added entries first,meaning that it behaves like a stack.(Contributed by Raymond Hettinger.)

• Therandom module'sRandom objects cannow be pickled on a 32-bit system and unpickled on a 64-bitsystem, and vice versa. Unfortunately, this change also meansthat Python 2.6'sRandom objects can't be unpickled correctlyon earlier versions of Python.(Contributed by Shawn Ligocki;bpo-1727780.)

  The newtriangular(low,high,mode) function returns randomnumbers following a triangular distribution. The returned valuesare betweenlow andhigh, not includinghigh itself, andwithmode as the most frequently occurring valuein the distribution. (Contributed by Wladmir van der Laan andRaymond Hettinger;bpo-1681432.)

• Long regular expression searches carried out by theremodule will check for signals being delivered, sotime-consuming searches can now be interrupted.(Contributed by Josh Hoyt and Ralf Schmitt;bpo-846388.)

  The regular expression module is implemented by compiling bytecodesfor a tiny regex-specific virtual machine. Untrusted codecould create malicious strings of bytecode directly and cause crashes,so Python 2.6 includes a verifier for the regex bytecode.(Contributed by Guido van Rossum from work for Google App Engine;bpo-3487.)

• Therlcompleter module'sCompleter.complete() methodwill now ignore exceptions triggered while evaluating a name.(Fixed by Lorenz Quack;bpo-2250.)

• Thesched module'sscheduler instances nowhave a read-onlyqueue attribute that returns thecontents of the scheduler's queue, represented as a list ofnamed tuples with the fields(time,priority,action,argument).(Contributed by Raymond Hettinger;bpo-1861.)

• Theselect module now has wrapper functionsfor the Linuxepoll() and BSDkqueue() system calls.modify() method was added to the existingpollobjects;pollobj.modify(fd,eventmask) takes a file descriptoror file object and an event mask, modifying the recorded event maskfor that file.(Contributed by Christian Heimes;bpo-1657.)

• Theshutil.copytree() function now has an optionalignore argumentthat takes a callable object. This callable will receive each directory pathand a list of the directory's contents, and returns a list of names thatwill be ignored, not copied.

  Theshutil module also provides anignore_patterns()function for use with this new parameter.ignore_patterns()takes an arbitrary number of glob-style patterns and returns acallable that will ignore any files and directories that match anyof these patterns. The following example copies a directory tree,but skips both.svn directories and Emacs backup files,which have names ending with '~':

  shutil.copytree('Doc/library','/tmp/library',ignore=shutil.ignore_patterns('*~','.svn'))

  （由 Tarek Ziadé 貢獻；bpo-2663。）

• Integrating signal handling with GUI handling event loopslike those used by Tkinter or GTk+ has long been a problem; mostsoftware ends up polling, waking up every fraction of a second to checkif any GUI events have occurred.Thesignal module can now make this more efficient.Callingsignal.set_wakeup_fd(fd) sets a file descriptorto be used; when a signal is received, a byte is written to thatfile descriptor. There's also a C-level function,PySignal_SetWakeupFd(), for setting the descriptor.

  Event loops will use this by opening a pipe to create two descriptors,one for reading and one for writing. The writable descriptorwill be passed toset_wakeup_fd(), and the readable descriptorwill be added to the list of descriptors monitored by the event loop viaselect() orpoll().On receiving a signal, a byte will be written and the main event loopwill be woken up, avoiding the need to poll.

  （由 Adam Olsen 貢獻；bpo-1583。）

  Thesiginterrupt() function is now available from Python code,and allows changing whether signals can interrupt system calls or not.(Contributed by Ralf Schmitt.)

  Thesetitimer() andgetitimer() functions have also beenadded (where they're available).setitimer()allows setting interval timers that will cause a signal to bedelivered to the process after a specified time, measured inwall-clock time, consumed process time, or combined process+systemtime. (Contributed by Guilherme Polo;bpo-2240.)

• Thesmtplib module now supports SMTP over SSL thanks to theaddition of theSMTP_SSL class. This class supports aninterface identical to the existingSMTP class.(Contributed by Monty Taylor.) Both class constructors also have anoptionaltimeout parameter that specifies a timeout for theinitial connection attempt, measured in seconds. (Contributed byFacundo Batista.)

  An implementation of the LMTP protocol (RFC 2033) was also addedto the module. LMTP is used in place of SMTP when transferringe-mail between agents that don't manage a mail queue. (LMTPimplemented by Leif Hedstrom;bpo-957003.)

  SMTP.starttls() now complies withRFC 3207 and forgets anyknowledge obtained from the server not obtained from the TLSnegotiation itself. (Patch contributed by Bill Fenner;bpo-829951.)

• Thesocket module now supports TIPC (https://tipc.sourceforge.net/),a high-performance non-IP-based protocol designed for use in clusteredenvironments. TIPC addresses are 4- or 5-tuples.(Contributed by Alberto Bertogli;bpo-1646.)

  A new function,create_connection(), takes an address andconnects to it using an optional timeout value, returning theconnected socket object. This function also looks up the address'stype and connects to it using IPv4 or IPv6 as appropriate. Changingyour code to usecreate_connection() instead ofsocket(socket.AF_INET,...) may be all that's required to makeyour code work with IPv6.

• The base classes in theSocketServer module now supportcalling ahandle_timeout() method after a span of inactivityspecified by the server'stimeout attribute. (Contributedby Michael Pomraning.) Theserve_forever() methodnow takes an optional poll interval measured in seconds,controlling how often the server will check for a shutdown request.(Contributed by Pedro Werneck and Jeffrey Yasskin;bpo-742598,bpo-1193577.)

• Thesqlite3 module, maintained by Gerhard Häring,has been updated from version 2.3.2 in Python 2.5 toversion 2.4.1.

• Thestruct module now supports the C99_Bool type,using the format character'?'.(Contributed by David Remahl.)

• ThePopen objects provided by thesubprocess modulenow haveterminate(),kill(), andsend_signal() methods.On Windows,send_signal() only supports theSIGTERMsignal, and all these methods are aliases for the Win32 API functionTerminateProcess().(Contributed by Christian Heimes.)

• A new variable in thesys module,float_info, is anobject containing information derived from thefloat.h fileabout the platform's floating-point support. Attributes of thisobject includemant_dig (number of digits in the mantissa),epsilon (smallest difference between 1.0 and the nextlargest value representable), and several others. (Contributed byChristian Heimes;bpo-1534.)

  Another new variable,dont_write_bytecode, controls whether Pythonwrites any.pyc or.pyo files on importing a module.If this variable is true, the compiled files are not written. Thevariable is initially set on start-up by supplying the-Bswitch to the Python interpreter, or by setting thePYTHONDONTWRITEBYTECODE environment variable beforerunning the interpreter. Python code can subsequentlychange the value of this variable to control whether bytecode filesare written or not.(Contributed by Neal Norwitz and Georg Brandl.)

  Information about the command-line arguments supplied to the Pythoninterpreter is available by reading attributes of a namedtuple available assys.flags. For example, theverboseattribute is true if Pythonwas executed in verbose mode,debug is true in debugging mode, etc.These attributes are all read-only.(Contributed by Christian Heimes.)

  A new function,getsizeof(), takes a Python object and returnsthe amount of memory used by the object, measured in bytes. Built-inobjects return correct results; third-party extensions may not,but can define a__sizeof__() method to return theobject's size.(Contributed by Robert Schuppenies;bpo-2898.)

  It's now possible to determine the current profiler and tracer functionsby callingsys.getprofile() andsys.gettrace().(Contributed by Georg Brandl;bpo-1648.)

• Thetarfile module now supports POSIX.1-2001 (pax) tarfiles inaddition to the POSIX.1-1988 (ustar) and GNU tar formats that werealready supported. The default format is GNU tar; specify theformat parameter to open a file using a different format:

  tar=tarfile.open("output.tar","w",format=tarfile.PAX_FORMAT)

  The newencoding anderrors parameters specify an encoding andan error handling scheme for character conversions.'strict','ignore', and'replace' are the three standard ways Python canhandle errors,;'utf-8' is a special value that replaces bad characters withtheir UTF-8 representation. (Character conversions occur because thePAX format supports Unicode filenames, defaulting to UTF-8 encoding.)

  TheTarFile.add() method now accepts anexclude argument that'sa function that can be used to exclude certain filenames froman archive.The function must take a filename and return true if the fileshould be excluded or false if it should be archived.The function is applied to both the name initially passed toadd()and to the names of files in recursively added directories.

  (All changes contributed by Lars Gustäbel).

• An optionaltimeout parameter was added to thetelnetlib.Telnet class constructor, specifying a timeoutmeasured in seconds. (Added by Facundo Batista.)

• Thetempfile.NamedTemporaryFile class usually deletesthe temporary file it created when the file is closed. Thisbehaviour can now be changed by passingdelete=False to theconstructor. (Contributed by Damien Miller;bpo-1537850.)

  A new class,SpooledTemporaryFile, behaves likea temporary file but stores its data in memory until a maximum size isexceeded. On reaching that limit, the contents will be written toan on-disk temporary file. (Contributed by Dustin J. Mitchell.)

  TheNamedTemporaryFile andSpooledTemporaryFile classesboth work as context managers, so you can writewithtempfile.NamedTemporaryFile()astmp:....(Contributed by Alexander Belopolsky;bpo-2021.)

• Thetest.test_support module gained a numberof context managers useful for writing tests.EnvironmentVarGuard() is acontext manager that temporarily changes environment variables andautomatically restores them to their old values.

  Another context manager,TransientResource, can surround callsto resources that may or may not be available; it will catch andignore a specified list of exceptions. For example,a network test may ignore certain failures when connecting to anexternal web site:

  withtest_support.TransientResource(IOError,errno=errno.ETIMEDOUT):f=urllib.urlopen('https://sf.net')...

  Finally,check_warnings() resets thewarning module'swarning filters and returns an object that will record all warningmessages triggered (bpo-3781):

  withtest_support.check_warnings()aswrec:warnings.simplefilter("always")# ... code that triggers a warning ...assertstr(wrec.message)=="function is outdated"assertlen(wrec.warnings)==1,"Multiple warnings raised"

  （由 Brett Cannon 貢獻。）

• Thetextwrap module can now preserve existing whitespaceat the beginnings and ends of the newly created linesby specifyingdrop_whitespace=Falseas an argument:

  >>>S="""This  sentence  has a bunch   of...  extra   whitespace.""">>>printtextwrap.fill(S,width=15)This  sentencehas a bunchof    extrawhitespace.>>>printtextwrap.fill(S,drop_whitespace=False,width=15)This  sentence  has a bunch   of    extra   whitespace.>>>

  （由 Dwayne Bailey 貢獻；bpo-1581073。）

• Thethreading module API is being changed to use propertiessuch asdaemon instead ofsetDaemon() andisDaemon() methods, and some methods have been renamed to useunderscores instead of camel-case; for example, theactiveCount() method is renamed toactive_count(). Boththe 2.6 and 3.0 versions of the module support the same propertiesand renamed methods, but don't remove the old methods. No date has been setfor the deprecation of the old APIs in Python 3.x; the old APIs won'tbe removed in any 2.x version.(Carried out by several people, most notably Benjamin Peterson.)

  Thethreading module'sThread objectsgained anident property that returns the thread'sidentifier, a nonzero integer. (Contributed by Gregory P. Smith;bpo-2871.)

• Thetimeit module now accepts callables as well as stringsfor the statement being timed and for the setup code.Two convenience functions were added for creatingTimer instances:repeat(stmt,setup,time,repeat,number) andtimeit(stmt,setup,time,number) create an instance and callthe corresponding method. (Contributed by Erik Demaine;bpo-1533909.)

• TheTkinter module now accepts lists and tuples for options,separating the elements by spaces before passing the resulting value toTcl/Tk.(Contributed by Guilherme Polo;bpo-2906.)

• Theturtle module for turtle graphics was greatly enhanced byGregor Lingl. New features in the module include:

  • Better animation of turtle movement and rotation.

  • Control over turtle movement using the newdelay(),tracer(), andspeed() methods.

  • The ability to set new shapes for the turtle, and todefine a new coordinate system.

  • Turtles now have anundo() method that can roll back actions.

  • Simple support for reacting to input events such as mouse and keyboardactivity, making it possible to write simple games.

  • Aturtle.cfg file can be used to customize the starting appearanceof the turtle's screen.

  • The module's docstrings can be replaced by new docstrings that have beentranslated into another language.

  (bpo-1513695)

• An optionaltimeout parameter was added to theurllib.urlopen function and theurllib.ftpwrapper class constructor, as well as theurllib2.urlopen function. The parameter specifies a timeoutmeasured in seconds. For example:

  >>>u=urllib2.urlopen("http://slow.example.com",                        timeout=3)Traceback (most recent call last):...urllib2.URLError:<urlopen error timed out>>>>

  (Added by Facundo Batista.)

• The Unicode database provided by theunicodedata modulehas been updated to version 5.1.0. (Updated byMartin von Löwis;bpo-3811.)

• Thewarnings module'sformatwarning() andshowwarning()gained an optionalline argument that can be used to supply theline of source code. (Added as part ofbpo-1631171, which re-implementedpart of thewarnings module in C code.)

  A new function,catch_warnings(), is a context managerintended for testing purposes that lets you temporarily modify thewarning filters and then restore their original values (bpo-3781).

• The XML-RPCSimpleXMLRPCServer andDocXMLRPCServerclasses can now be prevented from immediately opening and binding totheir socket by passingFalse as thebind_and_activateconstructor parameter. This can be used to modify the instance'sallow_reuse_address attribute before calling theserver_bind() andserver_activate() methods toopen the socket and begin listening for connections.(Contributed by Peter Parente;bpo-1599845.)

  SimpleXMLRPCServer also has a_send_traceback_headerattribute; if true, the exception and formatted traceback are returnedas HTTP headers "X-Exception" and "X-Traceback". This feature isfor debugging purposes only and should not be used on production serversbecause the tracebacks might reveal passwords or other sensitiveinformation. (Contributed by Alan McIntyre as part of hisproject for Google's Summer of Code 2007.)

• Thexmlrpclib module no longer automatically convertsdatetime.date anddatetime.time to thexmlrpclib.DateTime type; the conversion semantics werenot necessarily correct for all applications. Code usingxmlrpclib should convertdate andtimeinstances. (bpo-1330538) The code can also handledates before 1900 (contributed by Ralf Schmitt;bpo-2014)and 64-bit integers represented by using<i8> in XML-RPC responses(contributed by Riku Lindblad;bpo-2985).

• Thezipfile module'sZipFile class now hasextract() andextractall() methods that will unpacka single file or all the files in the archive to the current directory, orto a specified directory:

  z=zipfile.ZipFile('python-251.zip')# Unpack a single file, writing it relative# to the /tmp directory.z.extract('Python/sysmodule.c','/tmp')# Unpack all the files in the archive.z.extractall()

  （由 Alan McIntyre 貢獻；bpo-467924。）

  Theopen(),read() andextract() methods can nowtake either a filename or aZipInfo object. This is useful when anarchive accidentally contains a duplicated filename.(Contributed by Graham Horler;bpo-1775025.)

  Finally,zipfile now supports using Unicode filenamesfor archived files. (Contributed by Alexey Borzenkov;bpo-1734346.)

importastt=ast.parse("""d ={}for i in 'abcdefghijklm':    d[i + i] = ord(i) - ord('a') + 1print d""")printast.dump(t)

Module(body=[Assign(targets=[Name(id='d',ctx=Store())],value=Dict(keys=[],values=[]))For(target=Name(id='i',ctx=Store()),iter=Str(s='abcdefghijklm'),body=[Assign(targets=[Subscript(value=Name(id='d',ctx=Load()),slice=Index(value=BinOp(left=Name(id='i',ctx=Load()),op=Add(),right=Name(id='i',ctx=Load()))),ctx=Store())],value=BinOp(left=BinOp(left=Call(func=Name(id='ord',ctx=Load()),args=[Name(id='i',ctx=Load())],keywords=[],starargs=None,kwargs=None),op=Sub(),right=Call(func=Name(id='ord',ctx=Load()),args=[Str(s='a')],keywords=[],starargs=None,kwargs=None)),op=Add(),right=Num(n=1)))],orelse=[])Print(dest=None,values=[Name(id='d',ctx=Load())],nl=True)])

>>>literal='("a", "b", {2:4, 3:8, 1:2})'>>>printast.literal_eval(literal)('a', 'b', {1: 2, 2: 4, 3: 8})>>>printast.literal_eval('"a" + "b"')Traceback (most recent call last):...ValueError:malformed string

>>>importjson>>>data={"spam":"foo","parrot":42}>>>in_json=json.dumps(data)# Encode the data>>>in_json'{"parrot": 42, "spam": "foo"}'>>>json.loads(in_json)# Decode into a Python object{"spam": "foo", "parrot": 42}

importsysimportplistlibimportdatetime# Create data structuredata_struct=dict(lastAccessed=datetime.datetime.now(),version=1,categories=('Personal','Shared','Private'))# Create string containing XML.plist_str=plistlib.writePlistToString(data_struct)new_struct=plistlib.readPlistFromString(plist_str)printdata_structprintnew_struct# Write data structure to a file and read it back.plistlib.writePlist(data_struct,'/tmp/customizations.plist')new_struct=plistlib.readPlist('/tmp/customizations.plist')# read/writePlist accepts file-like objects as well as paths.plistlib.writePlist(data_struct,sys.stdout)

Deprecations and Removals¶

• String exceptions have been removed. Attempting to use them raises aTypeError.

• Changes to theException interfaceas dictated byPEP 352 continue to be made. For 2.6,themessage attribute is being deprecated in favor of theargs attribute.

• (3.0-warning mode) Python 3.0 will feature a reorganized standardlibrary that will drop many outdated modules and rename others.Python 2.6 running in 3.0-warning mode will warn about these moduleswhen they are imported.

  The list of deprecated modules is:audiodev,bgenlocations,buildtools,bundlebuilder,Canvas,compiler,dircache,dl,fpformat,gensuitemodule,ihooks,imageop,imgfile,linuxaudiodev,mhlib,mimetools,multifile,new,pure,statvfs,sunaudiodev,test.testall, andtoaiff.

• Thegopherlib module has been removed.

• TheMimeWriter module andmimify modulehave been deprecated; use theemailpackage instead.

• Themd5 module has been deprecated; use thehashlib moduleinstead.

• Theposixfile module has been deprecated;fcntl.lockf()provides better locking.

• Thepopen2 module has been deprecated; use thesubprocessmodule.

• Thergbimg module has been removed.

• Thesets module has been deprecated; it's better touse the built-inset andfrozenset types.

• Thesha module has been deprecated; use thehashlib moduleinstead.

建置和 C API 變更¶

Python 建置程序和 C API 的變更包括：

• Python now must be compiled with C89 compilers (after 19years!). This means that the Python source tree has dropped itsown implementations ofmemmove() andstrerror(), whichare in the C89 standard library.

• Python 2.6 can be built with Microsoft Visual Studio 2008 (version9.0), and this is the new default compiler. See thePCbuild directory for the build files. (Implemented byChristian Heimes.)

• On Mac OS X, Python 2.6 can be compiled as a 4-way universal build.Theconfigure scriptcan take a--with-universal-archs=[32-bit|64-bit|all]switch, controlling whether the binaries are built for 32-bitarchitectures (x86, PowerPC), 64-bit (x86-64 and PPC-64), or both.(Contributed by Ronald Oussoren.)

• A new function added in Python 2.6.6,PySys_SetArgvEx(), setsthe value ofsys.argv and can optionally updatesys.path toinclude the directory containing the script named bysys.argv[0]depending on the value of anupdatepath parameter.

  This function was added to close a security hole for applicationsthat embed Python. The old function,PySys_SetArgv(), wouldalways updatesys.path, and sometimes it would add the currentdirectory. This meant that, if you ran an application embeddingPython in a directory controlled by someone else, attackers couldput a Trojan-horse module in the directory (say, a file namedos.py) that your application would then import and run.

  If you maintain a C/C++ application that embeds Python, checkwhether you're callingPySys_SetArgv() and carefully considerwhether the application should be usingPySys_SetArgvEx()withupdatepath set to false. Note that using this function willbreak compatibility with Python versions 2.6.5 and earlier; if youhave to continue working with earlier versions, you can leavethe call toPySys_SetArgv() alone and callPyRun_SimpleString("sys.path.pop(0)\n") afterwards to discardthe firstsys.path component.

  Security issue reported asCVE 2008-5983;discussed ingh-50003, and fixed by Antoine Pitrou.

• The BerkeleyDB module now has a C API object, available asbsddb.db.api. This object can be used by other C extensionsthat wish to use thebsddb module for their own purposes.(Contributed by Duncan Grisby.)

• The new buffer interface, previously described inthe PEP 3118 section,addsPyObject_GetBuffer() andPyBuffer_Release(),as well as a few other functions.

• Python's use of the C stdio library is now thread-safe, or at leastas thread-safe as the underlying library is. A long-standing potentialbug occurred if one thread closed a file object while another threadwas reading from or writing to the object. In 2.6 file objectshave a reference count, manipulated by thePyFile_IncUseCount() andPyFile_DecUseCount()functions. File objects can't be closed unless the reference countis zero.PyFile_IncUseCount() should be called while the GILis still held, before carrying out an I/O operation using theFILE* pointer, andPyFile_DecUseCount() should be calledimmediately after the GIL is re-acquired.(Contributed by Antoine Pitrou and Gregory P. Smith.)

• Importing modules simultaneously in two different threads no longerdeadlocks; it will now raise anImportError. A new APIfunction,PyImport_ImportModuleNoBlock(), will look for amodule insys.modules first, then try to import it afteracquiring an import lock. If the import lock is held by anotherthread, anImportError is raised.(Contributed by Christian Heimes.)

• Several functions return information about the platform'sfloating-point support.PyFloat_GetMax() returnsthe maximum representable floating-point value,andPyFloat_GetMin() returns the minimumpositive value.PyFloat_GetInfo() returns an objectcontaining more information from thefloat.h file, such as"mant_dig" (number of digits in the mantissa),"epsilon"(smallest difference between 1.0 and the next largest valuerepresentable), and several others.(Contributed by Christian Heimes;bpo-1534.)

• C functions and methods that usePyComplex_AsCComplex() will now accept arguments thathave a__complex__() method. In particular, the functions in thecmath module will now accept objects with this method.This is a backport of a Python 3.0 change.(Contributed by Mark Dickinson;bpo-1675423.)

• Python's C API now includes two functions for case-insensitive stringcomparisons,PyOS_stricmp(char*,char*)andPyOS_strnicmp(char*,char*,Py_ssize_t).(Contributed by Christian Heimes;bpo-1635.)

• Many C extensions define their own little macro for addingintegers and strings to the module's dictionary in theinit* function. Python 2.6 finally defines standard macrosfor adding values to a module,PyModule_AddStringMacroandPyModule_AddIntMacro(). (Contributed byChristian Heimes.)

• Some macros were renamed in both 3.0 and 2.6 to make it clearer thatthey are macros,not functions.Py_Size() becamePy_SIZE(),Py_Type() becamePy_TYPE(), andPy_Refcnt() becamePy_REFCNT().The mixed-case macros are still availablein Python 2.6 for backward compatibility.(bpo-1629)

• Distutils now places C extensions it builds in adifferent directory when running on a debug version of Python.(Contributed by Collin Winter;bpo-1530959.)

• Several basic data types, such as integers and strings, maintaininternal free lists of objects that can be re-used. The datastructures for these free lists now follow a naming convention: thevariable is always namedfree_list, the counter is always namednumfree, and a macroPy<typename>_MAXFREELIST isalways defined.

• A new Makefile target, "make patchcheck", prepares the Python source treefor making a patch: it fixes trailing whitespace in all modified.py files, checks whether the documentation has been changed,and reports whether theMisc/ACKS andMisc/NEWS fileshave been updated.(Contributed by Brett Cannon.)

  Another new target, "make profile-opt", compiles a Python binaryusing GCC's profile-guided optimization. It compiles Python withprofiling enabled, runs the test suite to obtain a set of profilingresults, and then compiles using these results for optimization.(Contributed by Gregory P. Smith.)

Porting to Python 2.6¶

This section lists previously described changes and other bugfixesthat may require changes to your code:

• Classes that aren't supposed to be hashable shouldset__hash__=None in their definitions to indicatethe fact.

• String exceptions have been removed. Attempting to use them raises aTypeError.

• The__init__() method ofcollections.dequenow clears any existing contents of the dequebefore adding elements from the iterable. This change makes thebehavior matchlist.__init__().

• object.__init__() previously accepted arbitrary arguments andkeyword arguments, ignoring them. In Python 2.6, this is no longerallowed and will result in aTypeError. This will affect__init__() methods that end up calling the correspondingmethod onobject (perhaps through usingsuper()).Seebpo-1683368 for discussion.

• TheDecimal constructor now accepts leading and trailingwhitespace when passed a string. Previously it would raise anInvalidOperation exception. On the other hand, thecreate_decimal() method ofContext objects nowexplicitly disallows extra whitespace, raising aConversionSyntax exception.

• Due to an implementation accident, if you passed a file path tothe built-in__import__() function, it would actually importthe specified file. This was never intended to work, however, andthe implementation now explicitly checks for this case and raisesanImportError.

• C API: thePyImport_Import() andPyImport_ImportModule()functions now default to absolute imports, not relative imports.This will affect C extensions that import other modules.

• C API: extension data types that shouldn't be hashableshould define theirtp_hash slot toPyObject_HashNotImplemented().

• Thesocket module exceptionsocket.error now inheritsfromIOError. Previously it wasn't a subclass ofStandardError but now it is, throughIOError.(Implemented by Gregory P. Smith;bpo-1706815.)

• Thexmlrpclib module no longer automatically convertsdatetime.date anddatetime.time to thexmlrpclib.DateTime type; the conversion semantics werenot necessarily correct for all applications. Code usingxmlrpclib should convertdate andtimeinstances. (bpo-1330538)

• (3.0-warning mode) TheException class now warnswhen accessed using slicing or index access; havingException behave like a tuple is being phased out.

• (3.0-warning mode) inequality comparisons between two dictionariesor two objects that don't implement comparison methods are reportedas warnings.dict1==dict2 still works, butdict1<dict2is being phased out.

  Comparisons between cells, which are an implementation detail of Python'sscoping rules, also cause warnings because such comparisons are forbiddenentirely in 3.0.

For applications that embed Python:

• ThePySys_SetArgvEx() function was added in Python 2.6.6,letting applications close a security hole when the existingPySys_SetArgv() function was used. Check whether you'recallingPySys_SetArgv() and carefully consider whether theapplication should be usingPySys_SetArgvEx() withupdatepath set to false.

致謝¶

The author would like to thank the following people for offeringsuggestions, corrections and assistance with various drafts of thisarticle: Georg Brandl, Steve Brown, Nick Coghlan, Ralph Corderoy,Jim Jewett, Kent Johnson, Chris Lambacher, Martin Michlmayr,Antoine Pitrou, Brian Warner.