Editor:Raymond Hettinger
This article explains the new features in Python 3.8, compared to 3.7. Python 3.8 was released on October 14, 2019. For full details, see the
.
Summary – Release highlights
New Features
Assignment expressions
There is new syntax := that assigns values to variables as part of a larger expression. It is affectionately known as “the walrus operator” due to its resemblance to
the eyes and tusks of a walrus
.
In this example, the assignment expression helps avoid calling
twice:
if(n:=len(a))>10:print(f"List is too long ({n} elements, expected <= 10)")A similar benefit arises during regular expression matching where match objects are needed twice, once to test whether a match occurred and another to extract a subgroup:
discount=0.0if(mo:=re.search(r'(\d+)% discount',advertisement)):discount=float(mo.group(1))/100.0The operator is also useful with while-loops that compute a value to test loop termination and then need that same value again in the body of the loop:
# Loop over fixed length blockswhile(block:=f.read(256))!='':process(block)Another motivating use case arises in list comprehensions where a value computed in a filtering condition is also needed in the expression body:
[clean_name.title()fornameinnamesif(clean_name:=normalize('NFC',name))inallowed_names]Try to limit use of the walrus operator to clean cases that reduce complexity and improve readability.
See
for a full description.
(Contributed by Emily Morehouse in
.)
Positional-only parameters
There is a new function parameter syntax / to indicate that some function parameters must be specified positionally and cannot be used as keyword arguments. This is the same notation shown by help() for C functions annotated with Larry Hastings’
tool.
In the following example, parameters a and b are positional-only, while c or d can be positional or keyword, and e or f are required to be keywords:
deff(a,b,/,c,d,*,e,f):print(a,b,c,d,e,f)The following is a valid call:
f(10,20,30,d=40,e=50,f=60)However, these are invalid calls:
f(10,b=20,c=30,d=40,e=50,f=60)# b cannot be a keyword argumentf(10,20,30,40,50,f=60)# e must be a keyword argumentOne use case for this notation is that it allows pure Python functions to fully emulate behaviors of existing C coded functions. For example, the built-in
function does not accept keyword arguments:
defdivmod(a,b,/):"Emulate the built in divmod() function"return(a//b,a%b)Another use case is to preclude keyword arguments when the parameter name is not helpful. For example, the builtin
function has the signature len(obj,/). This precludes awkward calls such as:
len(obj='hello')# The "obj" keyword argument impairs readabilityA further benefit of marking a parameter as positional-only is that it allows the parameter name to be changed in the future without risk of breaking client code. For example, in the
module, the parameter name dist may be changed in the future. This was made possible with the following function specification:
defquantiles(dist,/,*,n=4,method='exclusive')...Since the parameters to the left of / are not exposed as possible keywords, the parameters names remain available for use in **kwargs:
>>> deff(a,b,/,**kwargs):... print(a,b,kwargs)...>>> f(10,20,a=1,b=2,c=3)# a and b are used in two ways10 20 {'a': 1, 'b': 2, 'c': 3}This greatly simplifies the implementation of functions and methods that need to accept arbitrary keyword arguments. For example, here is an excerpt from code in the
module:
classCounter(dict):def__init__(self,iterable=None,/,**kwds):# Note "iterable" is a possible keyword argumentSee
for a full description.
(Contributed by Pablo Galindo in
.)
Parallel filesystem cache for compiled bytecode files
The new
setting (also available as
pycache_prefix) configures the implicit bytecode cache to use a separate parallel filesystem tree, rather than the default __pycache__ subdirectories within each source directory.
The location of the cache is reported in
(
indicates the default location in __pycache__ subdirectories).
(Contributed by Carl Meyer in
.)
Debug build uses the same ABI as release build
The ABI of Python
is now compatible with Python release builds. On Unix, when Python is built in debug mode, it is now possible to load C extensions built in release mode and C extensions built using the stable ABI. The inverse is not true, as debug builds expose additional symbols not available in release builds.
Defining the Py_DEBUG macro no longer implies the Py_TRACE_REFS macro, which introduces the only ABI incompatibility. The Py_TRACE_REFS macro, which adds the
function and the
environment variable, can be set using the new
build option. (Contributed by Victor Stinner in
.)
On Unix, C extensions are no longer linked to libpython except on Android and Cygwin. It is now possible for a statically linked Python to load a C extension built using a shared library Python. (Contributed by Victor Stinner in
.)
On Unix, when Python is built in debug mode, import now also looks for C extensions compiled in release mode and for C extensions compiled with the stable ABI. (Contributed by Victor Stinner in
.)
To embed Python into an application, a new --embed option must be passed to python3-config--libs--embed to get -lpython3.8 (link the application to libpython). To support both 3.8 and older, try python3-config--libs--embed first and fallback to python3-config--libs (without --embed) if the previous command fails.
Add a pkg-config python-3.8-embed module to embed Python into an application: pkg-configpython-3.8-embed--libs includes -lpython3.8. To support both 3.8 and older, try pkg-configpython-X.Y-embed--libs first and fallback to pkg-configpython-X.Y--libs (without --embed) if the previous command fails (replace X.Y with the Python version).
On the other hand, pkg-configpython3.8--libs no longer contains -lpython3.8. C extensions must not be linked to libpython (except on Android and Cygwin, whose cases are handled by the script); this change is backward incompatible on purpose. (Contributed by Victor Stinner in
.)
f-strings support = for self-documenting expressions and debugging
Added an = specifier to
s. An f-string such as f'{expr=}' will expand to the text of the expression, an equal sign, then the representation of the evaluated expression. For example:
>>> importdatetimeasdt>>> user='eric_idle'>>> member_since=dt.date(1975,7,31)>>> f'{user=}{member_since=}'"user='eric_idle' member_since=datetime.date(1975, 7, 31)"The usual
allow more control over how the result of the expression is displayed:
>>> delta=dt.date.today()-member_since>>> f'{user=!s}{delta.days=:,d}''user=eric_idle delta.days=16,075'The = specifier will display the whole expression so that calculations can be shown:
>>> print(f'{theta=}{cos(radians(theta))=:.3f}')theta=30 cos(radians(theta))=0.866(Contributed by Eric V. Smith and Larry Hastings in
.)
PEP 578: Python Runtime Audit Hooks
The PEP adds an Audit Hook and Verified Open Hook. Both are available from Python and native code, allowing applications and frameworks written in pure Python code to take advantage of extra notifications, while also allowing embedders or system administrators to deploy builds of Python where auditing is always enabled.
See
for full details.
PEP 587: Python Initialization Configuration
The
adds a new C API to configure the Python Initialization providing finer control on the whole configuration and better error reporting.
New structures:
New functions:
PyPreConfig_InitIsolatedConfig()
PyPreConfig_InitPythonConfig()
Py_PreInitializeFromBytesArgs()
This PEP also adds _PyRuntimeState.preconfig (
type) and PyInterpreterState.config (
type) fields to these internal structures. PyInterpreterState.config becomes the new reference configuration, replacing global configuration variables and other private variables.
See
Python Initialization Configuration
for the documentation.
See
for a full description.
(Contributed by Victor Stinner in
.)
PEP 590: Vectorcall: a fast calling protocol for CPython
is added to the Python/C API. It is meant to formalize existing optimizations which were already done for various classes. Any
implementing a callable can use this protocol.
This is currently provisional. The aim is to make it fully public in Python 3.9.
See
for a full description.
(Contributed by Jeroen Demeyer, Mark Shannon and Petr Viktorin in
.)
Pickle protocol 5 with out-of-band data buffers
When
is used to transfer large data between Python processes in order to take advantage of multi-core or multi-machine processing, it is important to optimize the transfer by reducing memory copies, and possibly by applying custom techniques such as data-dependent compression.
The
protocol 5 introduces support for out-of-band buffers where
-compatible data can be transmitted separately from the main pickle stream, at the discretion of the communication layer.
See
for a full description.
(Contributed by Antoine Pitrou in
.)
Other Language Changes
A
statement was illegal in the
clause due to a problem with the implementation. In Python 3.8 this restriction was lifted. (Contributed by Serhiy Storchaka in
.)
The
,
, and
types now have an
method like that found in
and
. This minor API extension makes it possible to write numerator,denominator=x.as_integer_ratio() and have it work across multiple numeric types. (Contributed by Lisa Roach in
and Raymond Hettinger in
.)
Constructors of
,
and
will now use the
special method, if available and the corresponding method
,
or
is not available. (Contributed by Serhiy Storchaka in
.)
Added support of \N{name} escapes in
:
>>> notice='Copyright © 2019'>>> copyright_year_pattern=re.compile(r'\N{copyright sign}\s*(\d{4})')>>> int(copyright_year_pattern.search(notice).group(1))2019(Contributed by Jonathan Eunice and Serhiy Storchaka in
.)
Dict and dictviews are now iterable in reversed insertion order using
. (Contributed by Rémi Lapeyre in
.)
The syntax allowed for keyword names in function calls was further restricted. In particular, f((keyword)=arg) is no longer allowed. It was never intended to permit more than a bare name on the left-hand side of a keyword argument assignment term. (Contributed by Benjamin Peterson in
.)
Generalized iterable unpacking in
and
statements no longer requires enclosing parentheses. This brings the yield and return syntax into better agreement with normal assignment syntax:
>>> defparse(family):... lastname,*members=family.split()... returnlastname.upper(),*members...>>> parse('simpsons homer marge bart lisa maggie')('SIMPSONS', 'homer', 'marge', 'bart', 'lisa', 'maggie')(Contributed by David Cuthbert and Jordan Chapman in
.)
When a comma is missed in code such as [(10,20)(30,40)], the compiler displays a
with a helpful suggestion. This improves on just having a
indicating that the first tuple was not callable. (Contributed by Serhiy Storchaka in
.)
Arithmetic operations between subclasses of
or
and
objects now return an instance of the subclass, rather than the base class. This also affects the return type of operations whose implementation (directly or indirectly) uses datetime.timedelta arithmetic, such as
. (Contributed by Paul Ganssle in
.)
When the Python interpreter is interrupted by Ctrl-C (SIGINT) and the resulting
exception is not caught, the Python process now exits via a SIGINT signal or with the correct exit code such that the calling process can detect that it died due to a Ctrl-C. Shells on POSIX and Windows use this to properly terminate scripts in interactive sessions. (Contributed by Google via Gregory P. Smith in
.)
Some advanced styles of programming require updating the
object for an existing function. Since code objects are immutable, a new code object needs to be created, one that is modeled on the existing code object. With 19 parameters, this was somewhat tedious. Now, the new replace() method makes it possible to create a clone with a few altered parameters.
Here’s an example that alters the
function to prevent the data parameter from being used as a keyword argument:
>>> fromstatisticsimportmean>>> mean(data=[10,20,90])40>>> mean.__code__=mean.__code__.replace(co_posonlyargcount=1)>>> mean(data=[10,20,90])Traceback (most recent call last):...TypeError: mean() got some positional-only arguments passed as keyword arguments: 'data'(Contributed by Victor Stinner in
.)
For integers, the three-argument form of the
function now permits the exponent to be negative in the case where the base is relatively prime to the modulus. It then computes a modular inverse to the base when the exponent is -1, and a suitable power of that inverse for other negative exponents. For example, to compute the
modular multiplicative inverse
of 38 modulo 137, write:
>>> pow(38,-1,137)119>>> 119*38%1371Modular inverses arise in the solution of
. For example, to find integer solutions for 4258𝑥+147𝑦=369, first rewrite as 4258𝑥≡369(mod147) then solve:
>>> x=369*pow(4258,-1,147)%147>>> y=(4258*x-369)//-147>>> 4258*x+147*y369(Contributed by Mark Dickinson in
.)
Dict comprehensions have been synced-up with dict literals so that the key is computed first and the value second:
>>> # Dict comprehension>>> cast={input('role? '):input('actor? ')foriinrange(2)}role? King Arthuractor? Chapmanrole? Black Knightactor? Cleese>>> # Dict literal>>> cast={input('role? '):input('actor? ')}role? Sir Robinactor? Eric IdleThe guaranteed execution order is helpful with assignment expressions because variables assigned in the key expression will be available in the value expression:
>>> names=['Martin von Löwis','Łukasz Langa','Walter Dörwald']>>> {(n:=normalize('NFC',name)).casefold():nfornameinnames}{'martin von löwis': 'Martin von Löwis', 'łukasz langa': 'Łukasz Langa', 'walter dörwald': 'Walter Dörwald'}(Contributed by Jörn Heissler in
.)
The
method can now return a tuple from two to six elements long. Formerly, five was the limit. The new, optional sixth element is a callable with a (obj,state) signature. This allows the direct control over the state-updating behavior of a specific object. If not None, this callable will have priority over the object’s
method. (Contributed by Pierre Glaser and Olivier Grisel in
.)
New Modules
The new
module provides (provisional) support for reading metadata from third-party packages. For example, it can extract an installed package’s version number, list of entry points, and more:
>>> # Note following example requires that the popular "requests">>> # package has been installed.>>>>>> fromimportlib.metadataimportversion,requires,files>>> version('requests')'2.22.0'>>> list(requires('requests'))['chardet (<3.1.0,>=3.0.2)']>>> list(files('requests'))[:5][PackagePath('requests-2.22.0.dist-info/INSTALLER'), PackagePath('requests-2.22.0.dist-info/LICENSE'), PackagePath('requests-2.22.0.dist-info/METADATA'), PackagePath('requests-2.22.0.dist-info/RECORD'), PackagePath('requests-2.22.0.dist-info/WHEEL')](Contributed by Barry Warsaw and Jason R. Coombs in
.)
Improved Modules
ast
AST nodes now have end_lineno and end_col_offset attributes, which give the precise location of the end of the node. (This only applies to nodes that have lineno and col_offset attributes.)
New function
returns the source code for a specific AST node.
(Contributed by Ivan Levkivskyi in
.)
The
function has some new flags:
type_comments=True causes it to return the text of
and
type comments associated with certain AST nodes;
mode='func_type' can be used to parse
“signature type comments” (returned for function definition AST nodes);
feature_version=(3,N) allows specifying an earlier Python 3 version. For example, feature_version=(3,4) will treat
and
as non-reserved words.
(Contributed by Guido van Rossum in
.)
asyncio
has graduated from the provisional to stable API. This function can be used to execute a
and return the result while automatically managing the event loop. For example:
importasyncioasyncdefmain():awaitasyncio.sleep(0)return42asyncio.run(main())This is roughly equivalent to:
importasyncioasyncdefmain():awaitasyncio.sleep(0)return42loop=asyncio.new_event_loop()asyncio.set_event_loop(loop)try:loop.run_until_complete(main())finally:asyncio.set_event_loop(None)loop.close()The actual implementation is significantly more complex. Thus,
should be the preferred way of running asyncio programs.
(Contributed by Yury Selivanov in
.)
Running python-masyncio launches a natively async REPL. This allows rapid experimentation with code that has a top-level
. There is no longer a need to directly call asyncio.run() which would spawn a new event loop on every invocation:
$ python -m asyncio asyncio REPL 3.8.0 Use "await" directly instead of "asyncio.run()". Type "help", "copyright", "credits" or "license" for more information. >>> import asyncio >>> await asyncio.sleep(10, result='hello') hello (Contributed by Yury Selivanov in
.)
The exception
now inherits from
rather than
and no longer inherits from
concurrent.futures.CancelledError
. (Contributed by Yury Selivanov in
.)
On Windows, the default event loop is now
. (Contributed by Victor Stinner in
.)
now also supports UDP. (Contributed by Adam Meily and Andrew Svetlov in
.)
can now be interrupted by
(“CTRL+C”). (Contributed by Vladimir Matveev in
.)
Added
for getting the wrapped coroutine within an
. (Contributed by Alex Grönholm in
.)
Asyncio tasks can now be named, either by passing the name keyword argument to
or the
event loop method, or by calling the
method on the task object. The task name is visible in the repr() output of
and can also be retrieved using the
method. (Contributed by Alex Grönholm in
.)
Added support for
to
asyncio.loop.create_connection()
. To specify the behavior, two new parameters have been added: happy_eyeballs_delay and interleave. The Happy Eyeballs algorithm improves responsiveness in applications that support IPv4 and IPv6 by attempting to simultaneously connect using both. (Contributed by twisteroid ambassador in
.)
builtins
The
built-in has been improved to accept the ast.PyCF_ALLOW_TOP_LEVEL_AWAIT flag. With this new flag passed, compile() will allow top-level await, asyncfor and asyncwith constructs that are usually considered invalid syntax. Asynchronous code object marked with the CO_COROUTINE flag may then be returned. (Contributed by Matthias Bussonnier in
)
collections
The
method for
now returns a
instead of a
. This works because regular dicts have guaranteed ordering since Python 3.7. If the extra features of OrderedDict are required, the suggested remediation is to cast the result to the desired type: OrderedDict(nt._asdict()). (Contributed by Raymond Hettinger in
.)
cProfile
The
class can now be used as a context manager. Profile a block of code by running:
importcProfilewithcProfile.Profile()asprofiler:# code to be profiled...(Contributed by Scott Sanderson in
.)
csv
The
now returns instances of
instead of a
. The tool is now faster and uses less memory while still preserving the field order. (Contributed by Michael Selik in
.)
curses
Added a new variable holding structured version information for the underlying ncurses library:
. (Contributed by Serhiy Storchaka in
.)
ctypes
On Windows,
and subclasses now accept a winmode parameter to specify flags for the underlying LoadLibraryEx call. The default flags are set to only load DLL dependencies from trusted locations, including the path where the DLL is stored (if a full or partial path is used to load the initial DLL) and paths added by
. (Contributed by Steve Dower in
.)
datetime
Added new alternate constructors
datetime.date.fromisocalendar()
and
datetime.datetime.fromisocalendar()
, which construct
and
objects respectively from ISO year, week number, and weekday; these are the inverse of each class’s isocalendar method. (Contributed by Paul Ganssle in
.)
functools
can now be used as a straight decorator rather than as a function returning a decorator. So both of these are now supported:
@lru_cachedeff(x):...@lru_cache(maxsize=256)deff(x):...(Contributed by Raymond Hettinger in
.)
Added a new
decorator, for computed properties cached for the life of the instance.
importfunctoolsimportstatisticsclassDataset:def__init__(self,sequence_of_numbers):[email protected]_propertydefvariance(self):returnstatistics.variance(self.data)(Contributed by Carl Meyer in
)
Added a new
functools.singledispatchmethod()
decorator that converts methods into
using
:
fromfunctoolsimportsingledispatchmethodfromcontextlibimportsuppressclassTaskManager:def__init__(self,tasks):self.tasks=list(tasks)@singledispatchmethoddefdiscard(self,value):withsuppress(ValueError):self.tasks.remove(value)@discard.register(list)def_(self,tasks):targets=set(tasks)self.tasks=[xforxinself.tasksifxnotintargets](Contributed by Ethan Smith in
)
gc
can now receive an optional generation parameter indicating a generation to get objects from. (Contributed by Pablo Galindo in
.)
gettext
Added
and its variants. (Contributed by Franz Glasner, Éric Araujo, and Cheryl Sabella in
.)
gzip
Added the mtime parameter to
for reproducible output. (Contributed by Guo Ci Teo in
.)
A
exception is now raised instead of
for certain types of invalid or corrupt gzip files. (Contributed by Filip Gruszczyński, Michele Orrù, and Zackery Spytz in
.)
IDLE and idlelib
Output over N lines (50 by default) is squeezed down to a button. N can be changed in the PyShell section of the General page of the Settings dialog. Fewer, but possibly extra long, lines can be squeezed by right clicking on the output. Squeezed output can be expanded in place by double-clicking the button or into the clipboard or a separate window by right-clicking the button. (Contributed by Tal Einat in
.)
Add “Run Customized” to the Run menu to run a module with customized settings. Any command line arguments entered are added to sys.argv. They also re-appear in the box for the next customized run. One can also suppress the normal Shell main module restart. (Contributed by Cheryl Sabella, Terry Jan Reedy, and others in
and
.)
Added optional line numbers for IDLE editor windows. Windows open without line numbers unless set otherwise in the General tab of the configuration dialog. Line numbers for an existing window are shown and hidden in the Options menu. (Contributed by Tal Einat and Saimadhav Heblikar in
.)
OS native encoding is now used for converting between Python strings and Tcl objects. This allows IDLE to work with emoji and other non-BMP characters. These characters can be displayed or copied and pasted to or from the clipboard. Converting strings from Tcl to Python and back now never fails. (Many people worked on this for eight years but the problem was finally solved by Serhiy Storchaka in
.)
New in 3.8.1:
Add option to toggle cursor blink off. (Contributed by Zackery Spytz in
.)
Escape key now closes IDLE completion windows. (Contributed by Johnny Najera in
.)
The changes above have been backported to 3.7 maintenance releases.
Add keywords to module name completion list. (Contributed by Terry J. Reedy in
.)
inspect
The
function can now find docstrings for __slots__ if that attribute is a
where the values are docstrings. This provides documentation options similar to what we already have for
,
, and
:
classAudioClip:__slots__={'bit_rate':'expressed in kilohertz to one decimal place','duration':'in seconds, rounded up to an integer'}def__init__(self,bit_rate,duration):self.bit_rate=round(bit_rate/1000.0,1)self.duration=ceil(duration)(Contributed by Raymond Hettinger in
.)
io
In development mode (
env) and in
, the
finalizer now logs the exception if the close() method fails. The exception is ignored silently by default in release build. (Contributed by Victor Stinner in
.)
itertools
The
function added an option initial keyword argument to specify an initial value:
>>> fromitertoolsimportaccumulate>>> list(accumulate([10,5,30,15],initial=1000))[1000, 1010, 1015, 1045, 1060](Contributed by Lisa Roach in
.)
json.tool
Add option --json-lines to parse every input line as a separate JSON object. (Contributed by Weipeng Hong in
.)
logging
Added a force keyword argument to
. When set to true, any existing handlers attached to the root logger are removed and closed before carrying out the configuration specified by the other arguments.
This solves a long-standing problem. Once a logger or basicConfig() had been called, subsequent calls to basicConfig() were silently ignored. This made it difficult to update, experiment with, or teach the various logging configuration options using the interactive prompt or a Jupyter notebook.
(Suggested by Raymond Hettinger, implemented by Donghee Na, and reviewed by Vinay Sajip in
.)
math
Added new function
for computing Euclidean distance between two points. (Contributed by Raymond Hettinger in
.)
Expanded the
function to handle multiple dimensions. Formerly, it only supported the 2-D case. (Contributed by Raymond Hettinger in
.)
Added new function,
, as analogous function to
that returns the product of a ‘start’ value (default: 1) times an iterable of numbers:
>>> prior=0.8>>> likelihoods=[0.625,0.84,0.30]>>> math.prod(likelihoods,start=prior)0.126(Contributed by Pablo Galindo in
.)
Added two new combinatoric functions
and
:
>>> math.perm(10,3)# Permutations of 10 things taken 3 at a time720>>> math.comb(10,3)# Combinations of 10 things taken 3 at a time120(Contributed by Yash Aggarwal, Keller Fuchs, Serhiy Storchaka, and Raymond Hettinger in
,
, and
.)
Added a new function
for computing accurate integer square roots without conversion to floating point. The new function supports arbitrarily large integers. It is faster than floor(sqrt(n)) but slower than
:
>>> r=650320427>>> s=r**2>>> isqrt(s-1)# correct650320426>>> floor(sqrt(s-1))# incorrect650320427(Contributed by Mark Dickinson in
.)
The function
no longer accepts arguments that are not int-like. (Contributed by Pablo Galindo in
.)
mmap
The
class now has an
method to access the madvise() system call. (Contributed by Zackery Spytz in
.)
multiprocessing
Added new
module. (Contributed by Davin Potts in
.)
On macOS, the spawn start method is now used by default. (Contributed by Victor Stinner in
.)
os
Added new function
on Windows for providing additional search paths for native dependencies when importing extension modules or loading DLLs using
. (Contributed by Steve Dower in
.)
A new
function was added to wrap the memfd_create() syscall. (Contributed by Zackery Spytz and Christian Heimes in
.)
On Windows, much of the manual logic for handling reparse points (including symlinks and directory junctions) has been delegated to the operating system. Specifically,
will now traverse anything supported by the operating system, while
will only open reparse points that identify as “name surrogates” while others are opened as for os.stat(). In all cases,
will only have S_IFLNK set for symbolic links and not other kinds of reparse points. To identify other kinds of reparse point, check the new
attribute.
On Windows,
is now able to read directory junctions. Note that
will return False for directory junctions, and so code that checks islink first will continue to treat junctions as directories, while code that handles errors from os.readlink() may now treat junctions as links.
(Contributed by Steve Dower in
.)
os.path
functions that return a boolean result like
,
,
,
,
, and
now return False instead of raising
or its subclasses
and
for paths that contain characters or bytes unrepresentable at the OS level. (Contributed by Serhiy Storchaka in
.)
on Windows now prefers the USERPROFILE environment variable and does not use HOME, which is not normally set for regular user accounts. (Contributed by Anthony Sottile in
.)
on Windows no longer returns True for a link to a non-existent directory.
on Windows now resolves reparse points, including symlinks and directory junctions.
(Contributed by Steve Dower in
.)
pathlib
methods that return a boolean result like
,
,
,
,
,
,
,
,
now return False instead of raising
or its subclass
for paths that contain characters unrepresentable at the OS level. (Contributed by Serhiy Storchaka in
.)
Added pathlib.Path.link_to() which creates a hard link pointing to a path. (Contributed by Joannah Nanjekye in
) Note that link_to was deprecated in 3.10 and removed in 3.12 in favor of a hardlink_to method added in 3.10 which matches the semantics of the existing symlink_to method.
pickle
extensions subclassing the C-optimized
can now override the pickling logic of functions and classes by defining the special
method. (Contributed by Pierre Glaser and Olivier Grisel in
.)
plistlib
Added new
and enabled support for reading and writing NSKeyedArchiver-encoded binary plists. (Contributed by Jon Janzen in
.)
pprint
The
module added a sort_dicts parameter to several functions. By default, those functions continue to sort dictionaries before rendering or printing. However, if sort_dicts is set to false, the dictionaries retain the order that keys were inserted. This can be useful for comparison to JSON inputs during debugging.
In addition, there is a convenience new function,
that is like
but with sort_dicts defaulting to False:
>>> frompprintimportpprint,pp>>> d=dict(source='input.txt',operation='filter',destination='output.txt')>>> pp(d,width=40)# Original order{'source': 'input.txt', 'operation': 'filter', 'destination': 'output.txt'}>>> pprint(d,width=40)# Keys sorted alphabetically{'destination': 'output.txt', 'operation': 'filter', 'source': 'input.txt'}(Contributed by Rémi Lapeyre in
.)
py_compile
now supports silent mode. (Contributed by Joannah Nanjekye in
.)
shlex
The new
function acts as the inverse of
. (Contributed by Bo Bayles in
.)
shutil
now accepts a new dirs_exist_ok keyword argument. (Contributed by Josh Bronson in
.)
now defaults to the modern pax (POSIX.1-2001) format for new archives to improve portability and standards conformance, inherited from the corresponding change to the
module. (Contributed by C.A.M. Gerlach in
.)
on Windows now removes directory junctions without recursively removing their contents first. (Contributed by Steve Dower in
.)
socket
Added
and
convenience functions to automate the necessary tasks usually involved when creating a server socket, including accepting both IPv4 and IPv6 connections on the same socket. (Contributed by Giampaolo Rodolà in
.)
The
,
, and
functions have been implemented on Windows. (Contributed by Zackery Spytz in
.)
ssl
Added
to enable and
verify_client_post_handshake()
to initiate TLS 1.3 post-handshake authentication. (Contributed by Christian Heimes in
.)
statistics
Added
as a faster, floating-point variant of
. (Contributed by Raymond Hettinger and Steven D’Aprano in
.)
Added
(Contributed by Raymond Hettinger in
.)
Added
that returns a list of the most common values. (Contributed by Raymond Hettinger in
.)
Added
that divides data or a distribution in to equiprobable intervals (e.g. quartiles, deciles, or percentiles). (Contributed by Raymond Hettinger in
.)
Added
, a tool for creating and manipulating normal distributions of a random variable. (Contributed by Raymond Hettinger in
.)
>>> temperature_feb=NormalDist.from_samples([4,12,-3,2,7,14])>>> temperature_feb.mean6.0>>> temperature_feb.stdev6.356099432828281>>> temperature_feb.cdf(3)# Chance of being under 3 degrees0.3184678262814532>>> # Relative chance of being 7 degrees versus 10 degrees>>> temperature_feb.pdf(7)/temperature_feb.pdf(10)1.2039930378537762>>> el_niño=NormalDist(4,2.5)>>> temperature_feb+=el_niño# Add in a climate effect>>> temperature_febNormalDist(mu=10.0, sigma=6.830080526611674)>>> temperature_feb*(9/5)+32# Convert to FahrenheitNormalDist(mu=50.0, sigma=12.294144947901014)>>> temperature_feb.samples(3)# Generate random samples[7.672102882379219, 12.000027119750287, 4.647488369766392]sys
Add new
function which can be overridden to control how “unraisable exceptions” are handled. It is called when an exception has occurred but there is no way for Python to handle it. For example, when a destructor raises an exception or during garbage collection (
). (Contributed by Victor Stinner in
.)
tarfile
The
module now defaults to the modern pax (POSIX.1-2001) format for new archives, instead of the previous GNU-specific one. This improves cross-platform portability with a consistent encoding (UTF-8) in a standardized and extensible format, and offers several other benefits. (Contributed by C.A.M. Gerlach in
.)
threading
Add a new
function which handles uncaught
exception. It can be overridden to control how uncaught threading.Thread.run() exceptions are handled. (Contributed by Victor Stinner in
.)
Add a new
function and a
attribute to the
class. These return the native integral Thread ID of the current thread assigned by the kernel. This feature is only available on certain platforms, see get_native_id for more information. (Contributed by Jake Tesler in
.)
tokenize
The
module now implicitly emits a NEWLINE token when provided with input that does not have a trailing new line. This behavior now matches what the C tokenizer does internally. (Contributed by Ammar Askar in
.)
tkinter
Added methods selection_from(), selection_present(), selection_range() and selection_to() in the tkinter.Spinbox class. (Contributed by Juliette Monsel in
.)
Added method moveto() in the tkinter.Canvas class. (Contributed by Juliette Monsel in
.)
The tkinter.PhotoImage class now has transparency_get() and transparency_set() methods. (Contributed by Zackery Spytz in
.)
time
Added new clock
for macOS 10.12. (Contributed by Joannah Nanjekye in
.)
typing
The
module incorporates several new features:
A dictionary type with per-key types. See
and
. TypedDict uses only string keys. By default, every key is required to be present. Specify “total=False” to allow keys to be optional:
classLocation(TypedDict,total=False):lat_long:tuplegrid_square:strxy_coordinate:tuple
Literal types. See
and
. Literal types indicate that a parameter or return value is constrained to one or more specific literal values:
defget_status(port:int)->Literal['connected','disconnected']:...
“Final” variables, functions, methods and classes. See
,
and
. The final qualifier instructs a static type checker to restrict subclassing, overriding, or reassignment:
pi:Final[float]=3.1415926536
Protocol definitions. See
,
and
. Simple ABCs like
are now Protocol subclasses.
New protocol class
.
New functions
and
.
unicodedata
The
module has been upgraded to use the
release.
New function
can be used to verify a string is in a specific normal form, often much faster than by actually normalizing the string. (Contributed by Max Belanger, David Euresti, and Greg Price in
and
).
unittest
Added
to support an asynchronous version of
. Appropriate new assert functions for testing have been added as well. (Contributed by Lisa Roach in
).
Added
and
to unittest to support cleanups for
and
. (Contributed by Lisa Roach in
.)
Several mock assert functions now also print a list of actual calls upon failure. (Contributed by Petter Strandmark in
.)
module gained support for coroutines to be used as test cases with
unittest.IsolatedAsyncioTestCase
. (Contributed by Andrew Svetlov in
.)
Example:
importunittestclassTestRequest(unittest.IsolatedAsyncioTestCase):asyncdefasyncSetUp(self):self.connection=awaitAsyncConnection()asyncdeftest_get(self):response=awaitself.connection.get("https://example.com")self.assertEqual(response.status_code,200)asyncdefasyncTearDown(self):awaitself.connection.close()if__name__=="__main__":unittest.main()venv
now includes an Activate.ps1 script on all platforms for activating virtual environments under PowerShell Core 6.1. (Contributed by Brett Cannon in
.)
weakref
The proxy objects returned by
now support the matrix multiplication operators @ and @= in addition to the other numeric operators. (Contributed by Mark Dickinson in
.)
xml
As mitigation against DTD and external entity retrieval, the
and
modules no longer process external entities by default. (Contributed by Christian Heimes in
.)
The .find*() methods in the
module support wildcard searches like {*}tag which ignores the namespace and {namespace}* which returns all tags in the given namespace. (Contributed by Stefan Behnel in
.)
The
module provides a new function
that implements C14N 2.0. (Contributed by Stefan Behnel in
.)
The target object of
xml.etree.ElementTree.XMLParser
can receive namespace declaration events through the new callback methods start_ns() and end_ns(). Additionally, the
xml.etree.ElementTree.TreeBuilder
target can be configured to process events about comments and processing instructions to include them in the generated tree. (Contributed by Stefan Behnel in
and
.)
xmlrpc
now supports an optional headers keyword argument for a sequence of HTTP headers to be sent with each request. Among other things, this makes it possible to upgrade from default basic authentication to faster session authentication. (Contributed by Cédric Krier in
.)
Optimizations
The
module can now use the
function in some cases for better performance. Currently, it is only used on macOS and Linux (using glibc 2.24 or newer) if all these conditions are met:
close_fds is false;
preexec_fn, pass_fds, cwd and start_new_session parameters are not set;
the executable path contains a directory.
(Contributed by Joannah Nanjekye and Victor Stinner in
.)
,
,
,
and
use platform-specific “fast-copy” syscalls on Linux and macOS in order to copy the file more efficiently. “fast-copy” means that the copying operation occurs within the kernel, avoiding the use of userspace buffers in Python as in “outfd.write(infd.read())”. On Windows shutil.copyfile() uses a bigger default buffer size (1 MiB instead of 16 KiB) and a
-based variant of
is used. The speedup for copying a 512 MiB file within the same partition is about +26% on Linux, +50% on macOS and +40% on Windows. Also, much less CPU cycles are consumed. See
Platform-dependent efficient copy operations
section. (Contributed by Giampaolo Rodolà in
.)
uses
function and all copy functions depending from it use cached
values. The speedup for copying a directory with 8000 files is around +9% on Linux, +20% on Windows and +30% on a Windows SMB share. Also the number of os.stat() syscalls is reduced by 38% making shutil.copytree() especially faster on network filesystems. (Contributed by Giampaolo Rodolà in
.)
The default protocol in the
module is now Protocol 4, first introduced in Python 3.4. It offers better performance and smaller size compared to Protocol 3 available since Python 3.0.
Removed one
member from PyGC_Head. All GC tracked objects (e.g. tuple, list, dict) size is reduced 4 or 8 bytes. (Contributed by Inada Naoki in
.)
now uses __slots__ to reduce its memory footprint. (Contributed by Wouter Bolsterlee and Tal Einat in
)
Improved performance of
by 33%. Optimized argument handling and added a fast path for the common case of a single non-negative integer index into a tuple (which is the typical use case in the standard library). (Contributed by Raymond Hettinger in
.)
Sped-up field lookups in
. They are now more than two times faster, making them the fastest form of instance variable lookup in Python. (Contributed by Raymond Hettinger, Pablo Galindo, and Joe Jevnik, Serhiy Storchaka in
.)
The
constructor does not overallocate the internal item buffer if the input iterable has a known length (the input implements __len__). This makes the created list 12% smaller on average. (Contributed by Raymond Hettinger and Pablo Galindo in
.)
Doubled the speed of class variable writes. When a non-dunder attribute was updated, there was an unnecessary call to update slots. (Contributed by Stefan Behnel, Pablo Galindo Salgado, Raymond Hettinger, Neil Schemenauer, and Serhiy Storchaka in
.)
Reduced an overhead of converting arguments passed to many builtin functions and methods. This sped up calling some simple builtin functions and methods up to 20–50%. (Contributed by Serhiy Storchaka in
,
and
.)
LOAD_GLOBAL instruction now uses new “per opcode cache” mechanism. It is about 40% faster now. (Contributed by Yury Selivanov and Inada Naoki in
.)
Build and C API Changes
Default
became an empty string: the m flag for pymalloc became useless (builds with and without pymalloc are ABI compatible) and so has been removed. (Contributed by Victor Stinner in
.)
Example of changes:
Only python3.8 program is installed, python3.8m program is gone.
Only python3.8-config script is installed, python3.8m-config script is gone.
The m flag has been removed from the suffix of dynamic library filenames: extension modules in the standard library as well as those produced and installed by third-party packages, like those downloaded from PyPI. On Linux, for example, the Python 3.7 suffix .cpython-37m-x86_64-linux-gnu.so became .cpython-38-x86_64-linux-gnu.so in Python 3.8.
The header files have been reorganized to better separate the different kinds of APIs:
Include/*.h should be the portable public stable C API.
Include/cpython/*.h should be the unstable C API specific to CPython; public API, with some private API prefixed by _Py or _PY.
Include/internal/*.h is the private internal C API very specific to CPython. This API comes with no backward compatibility warranty and should not be used outside CPython. It is only exposed for very specific needs like debuggers and profiles which has to access to CPython internals without calling functions. This API is now installed by makeinstall.
(Contributed by Victor Stinner in
and
, work initiated by Eric Snow in Python 3.7.)
Some macros have been converted to static inline functions: parameter types and return type are well defined, they don’t have issues specific to macros, variables have a local scopes. Examples:
,
,
PyObject_INIT, PyObject_INIT_VAR
Private functions: _PyObject_GC_TRACK(), _PyObject_GC_UNTRACK(), _Py_Dealloc()
(Contributed by Victor Stinner in
.)
The PyByteArray_Init() and PyByteArray_Fini() functions have been removed. They did nothing since Python 2.7.4 and Python 3.2.0, were excluded from the limited API (stable ABI), and were not documented. (Contributed by Victor Stinner in
.)
The result of
is now of type constchar* rather of char*. (Contributed by Serhiy Storchaka in
.)
The duality of Modules/Setup.dist and Modules/Setup has been removed. Previously, when updating the CPython source tree, one had to manually copy Modules/Setup.dist (inside the source tree) to Modules/Setup (inside the build tree) in order to reflect any changes upstream. This was of a small benefit to packagers at the expense of a frequent annoyance to developers following CPython development, as forgetting to copy the file could produce build failures.
Now the build system always reads from Modules/Setup inside the source tree. People who want to customize that file are encouraged to maintain their changes in a git fork of CPython or as patch files, as they would do for any other change to the source tree.
(Contributed by Antoine Pitrou in
.)
Functions that convert Python number to C integer like
and argument parsing functions like
with integer converting format units like 'i' will now use the
special method instead of
, if available. The deprecation warning will be emitted for objects with the __int__() method but without the __index__() method (like
and
).
will now return 1 for objects implementing __index__().
,
and
also now use the __index__() method if available. (Contributed by Serhiy Storchaka in
and
.)
Heap-allocated type objects will now increase their reference count in
(and its parallel macro PyObject_INIT) instead of in
. Types that modify instance allocation or deallocation may need to be adjusted. (Contributed by Eddie Elizondo in
.)
The new function PyCode_NewWithPosOnlyArgs() allows to create code objects like PyCode_New(), but with an extra posonlyargcount parameter for indicating the number of positional-only arguments. (Contributed by Pablo Galindo in
.)
Py_SetPath() now sets
to the program full path (
) rather than to the program name (
). (Contributed by Victor Stinner in
.)
Deprecated
The distutils bdist_wininst command is now deprecated, use bdist_wheel (wheel packages) instead. (Contributed by Victor Stinner in
.)
Deprecated methods getchildren() and getiterator() in the
module now emit a
instead of
. They will be removed in Python 3.9. (Contributed by Serhiy Storchaka in
.)
Passing an object that is not an instance of
concurrent.futures.ThreadPoolExecutor
to
is deprecated and will be prohibited in Python 3.9. (Contributed by Elvis Pranskevichus in
.)
The
methods of
xml.dom.pulldom.DOMEventStream
,
and
have been deprecated.
Implementations of these methods have been ignoring their index parameter, and returning the next item instead. (Contributed by Berker Peksag in
.)
The
class has deprecated the _field_types attribute in favor of the __annotations__ attribute which has the same information. (Contributed by Raymond Hettinger in
.)
classes Num, Str, Bytes, NameConstant and Ellipsis are considered deprecated and will be removed in future Python versions.
should be used instead. (Contributed by Serhiy Storchaka in
.)
methods visit_Num(), visit_Str(), visit_Bytes(), visit_NameConstant() and visit_Ellipsis() are deprecated now and will not be called in future Python versions. Add the
method to handle all constant nodes. (Contributed by Serhiy Storchaka in
.)
The @asyncio.coroutine
is deprecated and will be removed in version 3.10. Instead of @asyncio.coroutine, use
instead. (Contributed by Andrew Svetlov in
.)
In
, the explicit passing of a loop argument has been deprecated and will be removed in version 3.10 for the following:
,
,
,
,
,
,
,
,
,
,
,
,
,
asyncio.create_subprocess_exec()
, and
asyncio.create_subprocess_shell()
.
The explicit passing of coroutine objects to
has been deprecated and will be removed in version 3.11. (Contributed by Yury Selivanov in
.)
The following functions and methods are deprecated in the
module: lgettext(), ldgettext(), lngettext() and ldngettext(). They return encoded bytes, and it’s possible that you will get unexpected Unicode-related exceptions if there are encoding problems with the translated strings. It’s much better to use alternatives which return Unicode strings in Python 3. These functions have been broken for a long time.
Function bind_textdomain_codeset(), methods NullTranslations.output_charset() and NullTranslations.set_output_charset(), and the codeset parameter of functions
and
are also deprecated, since they are only used for the l*gettext() functions. (Contributed by Serhiy Storchaka in
.)
The isAlive() method of
has been deprecated. (Contributed by Donghee Na in
.)
Many builtin and extension functions that take integer arguments will now emit a deprecation warning for
s,
s and any other objects that can be converted to integers only with a loss (e.g. that have the
method but do not have the
method). In future version they will be errors. (Contributed by Serhiy Storchaka in
.)
Deprecated passing the following arguments as keyword arguments:
func in
,
,
, cProfile.Profile.runcall(),
,
and
.
function in
unittest.TestCase.addCleanup()
.
fn in the
method of
concurrent.futures.ThreadPoolExecutor
and
concurrent.futures.ProcessPoolExecutor
.
callback in
contextlib.ExitStack.callback()
, contextlib.AsyncExitStack.callback() and
contextlib.AsyncExitStack.push_async_callback()
.
c and typeid in the create() method of multiprocessing.managers.Server and multiprocessing.managers.SharedMemoryServer.
obj in
.
In future releases of Python, they will be
. (Contributed by Serhiy Storchaka in
.)
API and Feature Removals
The following features and APIs have been removed from Python 3.8:
Starting with Python 3.3, importing ABCs from
was deprecated, and importing should be done from
. Being able to import from collections was marked for removal in 3.8, but has been delayed to 3.9. (See
.)
The macpath module, deprecated in Python 3.7, has been removed. (Contributed by Victor Stinner in
.)
The function platform.popen() has been removed, after having been deprecated since Python 3.3: use
instead. (Contributed by Victor Stinner in
.)
The function time.clock() has been removed, after having been deprecated since Python 3.3: use
or
instead, depending on your requirements, to have well-defined behavior. (Contributed by Matthias Bussonnier in
.)
The pyvenv script has been removed in favor of python3.8-mvenv to help eliminate confusion as to what Python interpreter the pyvenv script is tied to. (Contributed by Brett Cannon in
.)
parse_qs, parse_qsl, and escape are removed from the cgi module. They are deprecated in Python 3.2 or older. They should be imported from the urllib.parse and html modules instead.
filemode function is removed from the
module. It is not documented and deprecated since Python 3.3.
The
constructor no longer accepts the html argument. It never had an effect and was deprecated in Python 3.4. All other parameters are now
. (Contributed by Serhiy Storchaka in
.)
Removed the doctype() method of
. (Contributed by Serhiy Storchaka in
.)
“unicode_internal” codec is removed. (Contributed by Inada Naoki in
.)
The Cache and Statement objects of the
module are not exposed to the user. (Contributed by Aviv Palivoda in
.)
The bufsize keyword argument of
and
which was ignored and deprecated since Python 3.6 has been removed.
(Contributed by Matthias Bussonnier.)
The functions sys.set_coroutine_wrapper() and sys.get_coroutine_wrapper() deprecated in Python 3.7 have been removed;
(Contributed by Matthias Bussonnier.)
Porting to Python 3.8
This section lists previously described changes and other bugfixes that may require changes to your code.
Changes in Python behavior
Yield expressions (both yield and yieldfrom clauses) are now disallowed in comprehensions and generator expressions (aside from the iterable expression in the leftmost for clause). (Contributed by Serhiy Storchaka in
.)
The compiler now produces a
when identity checks (is and isnot) are used with certain types of literals (e.g. strings, numbers). These can often work by accident in CPython, but are not guaranteed by the language spec. The warning advises users to use equality tests (== and !=) instead. (Contributed by Serhiy Storchaka in
.)
The CPython interpreter can swallow exceptions in some circumstances. In Python 3.8 this happens in fewer cases. In particular, exceptions raised when getting the attribute from the type dictionary are no longer ignored. (Contributed by Serhiy Storchaka in
.)
Removed __str__ implementations from builtin types
,
,
,
and few classes from the standard library. They now inherit __str__() from
. As result, defining the __repr__() method in the subclass of these classes will affect their string representation. (Contributed by Serhiy Storchaka in
.)
On AIX,
doesn’t contain the major version anymore. It is always 'aix', instead of 'aix3' .. 'aix7'. Since older Python versions include the version number, so it is recommended to always use sys.platform.startswith('aix'). (Contributed by M. Felt in
.)
PyEval_AcquireLock() and PyEval_AcquireThread() now terminate the current thread if called while the interpreter is finalizing, making them consistent with
,
, and
. If this behavior is not desired, guard the call by checking _Py_IsFinalizing() or
. (Contributed by Joannah Nanjekye in
.)
Changes in the Python API
The
function now uses the UTF-8 encoding on Windows, rather than the ANSI code page: see
for the rationale. The function is no longer deprecated on Windows. (Contributed by Victor Stinner in
.)
can now use
in some cases for better performance. On Windows Subsystem for Linux and QEMU User Emulation, the Popen constructor using os.posix_spawn() no longer raises an exception on errors like “missing program”. Instead the child process fails with a non-zero
. (Contributed by Joannah Nanjekye and Victor Stinner in
.)
The preexec_fn argument of *
is no longer compatible with subinterpreters. The use of the parameter in a subinterpreter now raises
. (Contributed by Eric Snow in
, modified by Christian Heimes in
.)
The
method no longer silently ignores arbitrary exceptions. (Contributed by Victor Stinner in
.)
The function platform.popen() has been removed, after having been deprecated since Python 3.3: use
instead. (Contributed by Victor Stinner in
.)
The
function no longer raises an exception when given multimodal data. Instead, it returns the first mode encountered in the input data. (Contributed by Raymond Hettinger in
.)
The
method of the
class no longer takes arguments. Using it with arguments for changing the selection was deprecated in Python 3.6. Use specialized methods like
for changing the selection. (Contributed by Serhiy Storchaka in
.)
The
,
and
methods of
and the
method of
now preserve the attribute order specified by the user. (Contributed by Diego Rojas and Raymond Hettinger in
.)
A
database opened with flags 'r' is now read-only.
with flags 'r' and 'w' no longer creates a database if it does not exist. (Contributed by Serhiy Storchaka in
.)
The doctype() method defined in a subclass of
will no longer be called and will emit a
instead of a
. Define the
method on a target for handling an XML doctype declaration. (Contributed by Serhiy Storchaka in
.)
A
is now raised when the custom metaclass doesn’t provide the __classcell__ entry in the namespace passed to type.__new__. A
was emitted in Python 3.6–3.7. (Contributed by Serhiy Storchaka in
.)
The
class can now be used as a context manager. (Contributed by Scott Sanderson in
.)
,
,
,
and
use platform-specific “fast-copy” syscalls (see
Platform-dependent efficient copy operations
section).
default buffer size on Windows was changed from 16 KiB to 1 MiB.
The PyGC_Head struct has changed completely. All code that touched the struct member should be rewritten. (See
.)
The
struct has been moved into the “internal” header files (specifically Include/internal/pycore_pystate.h). An opaque PyInterpreterState is still available as part of the public API (and stable ABI). The docs indicate that none of the struct’s fields are public, so we hope no one has been using them. However, if you do rely on one or more of those private fields and have no alternative then please open a BPO issue. We’ll work on helping you adjust (possibly including adding accessor functions to the public API). (See
.)
The
method now returns None on success and raises an exception on error under all platforms. Previously, its behavior was platform-dependent: a nonzero value was returned on success; zero was returned on error under Windows. A zero value was returned on success; an exception was raised on error under Unix. (Contributed by Berker Peksag in
.)
and
modules no longer process external entities by default. (Contributed by Christian Heimes in
.)
Deleting a key from a read-only
database (
,
or
) raises error (
,
or
) instead of
. (Contributed by Xiang Zhang in
.)
Simplified AST for literals. All constants will be represented as
instances. Instantiating old classes Num, Str, Bytes, NameConstant and Ellipsis will return an instance of Constant. (Contributed by Serhiy Storchaka in
.)
on Windows now prefers the USERPROFILE environment variable and does not use HOME, which is not normally set for regular user accounts. (Contributed by Anthony Sottile in
.)
The exception
now inherits from
rather than
and no longer inherits from
concurrent.futures.CancelledError
. (Contributed by Yury Selivanov in
.)
The function
now correctly waits for cancellation when using an instance of
. Previously, upon reaching timeout, it was cancelled and immediately returned. (Contributed by Elvis Pranskevichus in
.)
The function
asyncio.BaseTransport.get_extra_info()
now returns a safe to use socket object when ‘socket’ is passed to the name parameter. (Contributed by Yury Selivanov in
.)
has graduated to the stable API.
DLL dependencies for extension modules and DLLs loaded with
on Windows are now resolved more securely. Only the system paths, the directory containing the DLL or PYD file, and directories added with
are searched for load-time dependencies. Specifically, PATH and the current working directory are no longer used, and modifications to these will no longer have any effect on normal DLL resolution. If your application relies on these mechanisms, you should check for add_dll_directory() and if it exists, use it to add your DLLs directory while loading your library. Note that Windows 7 users will need to ensure that Windows Update KB2533623 has been installed (this is also verified by the installer). (Contributed by Steve Dower in
.)
The header files and functions related to pgen have been removed after its replacement by a pure Python implementation. (Contributed by Pablo Galindo in
.)
has a new parameter in the second position of the constructor (posonlyargcount) to support positional-only arguments defined in
. The first argument (argcount) now represents the total number of positional arguments (including positional-only arguments). The new replace() method of types.CodeType can be used to make the code future-proof.
The parameter digestmod for
no longer uses the MD5 digest by default.
Changes in the C API
The
structure got a new cf_feature_version field. It should be initialized to PY_MINOR_VERSION. The field is ignored by default, and is used if and only if PyCF_ONLY_AST flag is set in cf_flags. (Contributed by Guido van Rossum in
.)
The PyEval_ReInitThreads() function has been removed from the C API. It should not be called explicitly: use
instead. (Contributed by Victor Stinner in
.)
On Unix, C extensions are no longer linked to libpython except on Android and Cygwin. When Python is embedded, libpython must not be loaded with RTLD_LOCAL, but RTLD_GLOBAL instead. Previously, using RTLD_LOCAL, it was already not possible to load C extensions which were not linked to libpython, like C extensions of the standard library built by the *shared* section of Modules/Setup. (Contributed by Victor Stinner in
.)
Use of # variants of formats in parsing or building value (e.g.
,
,
, etc.) without PY_SSIZE_T_CLEAN defined raises DeprecationWarning now. It will be removed in 3.10 or 4.0. Read
Parsing arguments and building values
for detail. (Contributed by Inada Naoki in
.)
Instances of heap-allocated types (such as those created with
) hold a reference to their type object. Increasing the reference count of these type objects has been moved from
to the more low-level functions,
and PyObject_INIT. This makes types created through PyType_FromSpec() behave like other classes in managed code.
are not affected.
For the vast majority of cases, there should be no side effect. However, types that manually increase the reference count after allocating an instance (perhaps to work around the bug) may now become immortal. To avoid this, these classes need to call Py_DECREF on the type object during instance deallocation.
To correctly port these types into 3.8, please apply the following changes:
Remove
on the type object after allocating an instance - if any. This may happen after calling
,
,
,
, or any other custom allocator that uses
or PyObject_INIT.
Example:
staticfoo_struct*foo_new(PyObject*type){foo_struct*foo=PyObject_GC_New(foo_struct,(PyTypeObject*)type);if(foo==NULL)returnNULL;#if PY_VERSION_HEX < 0x03080000// Workaround for Python issue 35810; no longer necessary in Python 3.8PY_INCREF(type)#endifreturnfoo;}
Ensure that all custom tp_dealloc functions of heap-allocated types decrease the type’s reference count.
Example:
staticvoidfoo_dealloc(foo_struct*instance){PyObject*type=Py_TYPE(instance);PyObject_GC_Del(instance);#if PY_VERSION_HEX >= 0x03080000// This was not needed before Python 3.8 (Python issue 35810)Py_DECREF(type);#endif}
(Contributed by Eddie Elizondo in
.)
The
macro has been implemented for MSVC. The macro now must be placed before the symbol name.
Example:
Py_DEPRECATED(3.8)PyAPI_FUNC(int)Py_OldFunction(void);(Contributed by Zackery Spytz in
.)
The interpreter does not pretend to support binary compatibility of extension types across feature releases, anymore. A
exported by a third-party extension module is supposed to have all the slots expected in the current Python version, including
(
is not checked anymore before reading tp_finalize).
(Contributed by Antoine Pitrou in
.)
The functions PyNode_AddChild() and PyParser_AddToken() now accept two additional int arguments end_lineno and end_col_offset.
The libpython38.a file to allow MinGW tools to link directly against python38.dll is no longer included in the regular Windows distribution. If you require this file, it may be generated with the gendef and dlltool tools, which are part of the MinGW binutils package:
gendef-python38.dll>tmp.def dlltool--dllnamepython38.dll--deftmp.def--output-liblibpython38.a The location of an installed pythonXY.dll will depend on the installation options and the version and language of Windows. See
for more information. The resulting library should be placed in the same directory as pythonXY.lib, which is generally the libs directory under your Python installation.
(Contributed by Steve Dower in
.)
CPython bytecode changes
The interpreter loop has been simplified by moving the logic of unrolling the stack of blocks into the compiler. The compiler emits now explicit instructions for adjusting the stack of values and calling the cleaning-up code for
,
and
.
Removed opcodes BREAK_LOOP, CONTINUE_LOOP, SETUP_LOOP and SETUP_EXCEPT. Added new opcodes ROT_FOUR, BEGIN_FINALLY, CALL_FINALLY and POP_FINALLY. Changed the behavior of END_FINALLY and WITH_CLEANUP_START.
(Contributed by Mark Shannon, Antoine Pitrou and Serhiy Storchaka in
.)
Added new opcode
for handling exceptions raised when awaiting a next item in an
loop. (Contributed by Serhiy Storchaka in
.)
The
now expects the value as the first element in the stack and the key as the second element. This change was made so the key is always evaluated before the value in dictionary comprehensions, as proposed by
. (Contributed by Jörn Heissler in
.)
Demos and Tools
Added a benchmark script for timing various ways to access variables: Tools/scripts/var_access_benchmark.py. (Contributed by Raymond Hettinger in
.)
Here’s a summary of performance improvements since Python 3.3:
Python version 3.3 3.4 3.5 3.6 3.7 3.8 -------------- --- --- --- --- --- --- Variable and attribute read access: read_local 4.0 7.1 7.1 5.4 5.1 3.9 read_nonlocal 5.3 7.1 8.1 5.8 5.4 4.4 read_global 13.3 15.5 19.0 14.3 13.6 7.6 read_builtin 20.0 21.1 21.6 18.5 19.0 7.5 read_classvar_from_class 20.5 25.6 26.5 20.7 19.5 18.4 read_classvar_from_instance 18.5 22.8 23.5 18.8 17.1 16.4 read_instancevar 26.8 32.4 33.1 28.0 26.3 25.4 read_instancevar_slots 23.7 27.8 31.3 20.8 20.8 20.2 read_namedtuple 68.5 73.8 57.5 45.0 46.8 18.4 read_boundmethod 29.8 37.6 37.9 29.6 26.9 27.7 Variable and attribute write access: write_local 4.6 8.7 9.3 5.5 5.3 4.3 write_nonlocal 7.3 10.5 11.1 5.6 5.5 4.7 write_global 15.9 19.7 21.2 18.0 18.0 15.8 write_classvar 81.9 92.9 96.0 104.6 102.1 39.2 write_instancevar 36.4 44.6 45.8 40.0 38.9 35.5 write_instancevar_slots 28.7 35.6 36.1 27.3 26.6 25.7 Data structure read access: read_list 19.2 24.2 24.5 20.8 20.8 19.0 read_deque 19.9 24.7 25.5 20.2 20.6 19.8 read_dict 19.7 24.3 25.7 22.3 23.0 21.0 read_strdict 17.9 22.6 24.3 19.5 21.2 18.9 Data structure write access: write_list 21.2 27.1 28.5 22.5 21.6 20.0 write_deque 23.8 28.7 30.1 22.7 21.8 23.5 write_dict 25.9 31.4 33.3 29.3 29.2 24.7 write_strdict 22.9 28.4 29.9 27.5 25.2 23.1 Stack (or queue) operations: list_append_pop 144.2 93.4 112.7 75.4 74.2 50.8 deque_append_pop 30.4 43.5 57.0 49.4 49.2 42.5 deque_append_popleft 30.8 43.7 57.3 49.7 49.7 42.8 Timing loop: loop_overhead 0.3 0.5 0.6 0.4 0.3 0.3 The benchmarks were measured on an
Intel® Core™ i7-4960HQ processor
running the macOS 64-bit builds found at
. The benchmark script displays timings in nanoseconds.
Notable changes in Python 3.8.1
Due to significant security concerns, the reuse_address parameter of
asyncio.loop.create_datagram_endpoint()
is no longer supported. This is because of the behavior of the socket option SO_REUSEADDR in UDP. For more details, see the documentation for loop.create_datagram_endpoint(). (Contributed by Kyle Stanley, Antoine Pitrou, and Yury Selivanov in
.)
Notable changes in Python 3.8.2
Fixed a regression with the ignore callback of
. The argument types are now str and List[str] again. (Contributed by Manuel Barkhau and Giampaolo Rodola in
.)
Notable changes in Python 3.8.3
The constant values of future flags in the
module are updated in order to prevent collision with compiler flags. Previously PyCF_ALLOW_TOP_LEVEL_AWAIT was clashing with CO_FUTURE_DIVISION. (Contributed by Batuhan Taskaya in
)
Notable changes in Python 3.8.8
Earlier Python versions allowed using both ; and & as query parameter separators in
and
. Due to security concerns, and to conform with newer W3C recommendations, this has been changed to allow only a single separator key, with & as the default. This change also affects cgi.parse() and cgi.parse_multipart() as they use the affected functions internally. For more details, please see their respective documentation. (Contributed by Adam Goldschmidt, Senthil Kumaran and Ken Jin in
.)
Notable changes in Python 3.8.9
A security fix alters the
behavior to not trust the IPv4 address sent from the remote server when setting up a passive data channel. We reuse the ftp server IP address instead. For unusual code requiring the old behavior, set a trust_server_pasv_ipv4_address attribute on your FTP instance to True. (See
)
Notable changes in Python 3.8.10
macOS 11.0 (Big Sur) and Apple Silicon Mac support
As of 3.8.10, Python now supports building and running on macOS 11 (Big Sur) and on Apple Silicon Macs (based on the ARM64 architecture). A new universal build variant, universal2, is now available to natively support both ARM64 and Intel64 in one set of executables. Note that support for “weaklinking”, building binaries targeted for newer versions of macOS that will also run correctly on older versions by testing at runtime for missing features, is not included in this backport from Python 3.9; to support a range of macOS versions, continue to target for and build on the oldest version in the range.
(Originally contributed by Ronald Oussoren and Lawrence D’Anna in
, with fixes by FX Coudert and Eli Rykoff, and backported to 3.8 by Maxime Bélanger and Ned Deily)
Notable changes in Python 3.8.10
urllib.parse
The presence of newline or tab characters in parts of a URL allows for some forms of attacks. Following the WHATWG specification that updates
, ASCII newline \n, \r and tab \t characters are stripped from the URL by the parser in
preventing such attacks. The removal characters are controlled by a new module level variable urllib.parse._UNSAFE_URL_BYTES_TO_REMOVE. (See
)
Notable changes in Python 3.8.12
Changes in the Python API
Starting with Python 3.8.12 the
module no longer accepts any leading zeros in IPv4 address strings. Leading zeros are ambiguous and interpreted as octal notation by some libraries. For example the legacy function
treats leading zeros as octal notation. glibc implementation of modern
does not accept any leading zeros.
(Originally contributed by Christian Heimes in
, and backported to 3.8 by Achraf Merzouki.)
Notable security feature in 3.8.14
Converting between
and
in bases other than 2 (binary), 4, 8 (octal), 16 (hexadecimal), or 32 such as base 10 (decimal) now raises a
if the number of digits in string form is above a limit to avoid potential denial of service attacks due to the algorithmic complexity. This is a mitigation for
. This limit can be configured or disabled by environment variable, command line flag, or
APIs. See the
integer string conversion length limitation
documentation. The default limit is 4300 digits in string form.
Notable changes in 3.8.17
tarfile
The extraction methods in
, and
, have a new a filter argument that allows limiting tar features than may be surprising or dangerous, such as creating files outside the destination directory. See
for details. In Python 3.12, use without the filter argument will show a
. In Python 3.14, the default will switch to 'data'. (Contributed by Petr Viktorin in
.)
