What’s New In Python 3.11#
- Release
3.12.0a0
- Date
September 26, 2022
This article explains the new features in Python 3.11, compared to 3.10.
For full details, see the changelog.
Summary – Release highlights#
Python 3.11 is between 10-60% faster than Python 3.10. On average, we measured a 1.25x speedup on the standard benchmark suite. See Faster CPython for details.
New syntax features:
PEP 654: Exception Groups and
except*
.
New built-in features:
PEP 678: Enriching Exceptions with Notes.
New standard library modules:
PEP 680:
tomllib
— Support for Parsing TOML in the Standard Library.
Interpreter improvements:
PEP 657: Include Fine Grained Error Locations in Tracebacks.
New
-P
command line option andPYTHONSAFEPATH
environment variable to disable automatically prepending a potentially unsafe path (the working dir or script directory, depending on invocation) tosys.path
.
New typing features:
PEP 646: Variadic generics.
PEP 655: Marking individual TypedDict items as required or potentially missing.
PEP 673:
Self
type.PEP 675: Arbitrary literal string type.
PEP 681: Data Class Transforms.
Important deprecations, removals or restrictions:
New Features#
PEP 657: Enhanced error locations in tracebacks#
When printing tracebacks, the interpreter will now point to the exact expression that caused the error, instead of just the line. For example:
Traceback (most recent call last):
File "distance.py", line 11, in <module>
print(manhattan_distance(p1, p2))
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "distance.py", line 6, in manhattan_distance
return abs(point_1.x - point_2.x) + abs(point_1.y - point_2.y)
^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'x'
Previous versions of the interpreter would point to just the line, making it
ambiguous which object was None
. These enhanced errors can also be helpful
when dealing with deeply nested dict
objects and multiple function calls:
Traceback (most recent call last):
File "query.py", line 37, in <module>
magic_arithmetic('foo')
File "query.py", line 18, in magic_arithmetic
return add_counts(x) / 25
^^^^^^^^^^^^^
File "query.py", line 24, in add_counts
return 25 + query_user(user1) + query_user(user2)
^^^^^^^^^^^^^^^^^
File "query.py", line 32, in query_user
return 1 + query_count(db, response['a']['b']['c']['user'], retry=True)
~~~~~~~~~~~~~~~~~~^^^^^
TypeError: 'NoneType' object is not subscriptable
As well as complex arithmetic expressions:
Traceback (most recent call last):
File "calculation.py", line 54, in <module>
result = (x / y / z) * (a / b / c)
~~~~~~^~~
ZeroDivisionError: division by zero
Additionally, the information used by the enhanced traceback feature is made available via a general API, that can be used to correlate bytecode instructions with source code location. This information can be retrieved using:
The
codeobject.co_positions()
method in Python.The
PyCode_Addr2Location()
function in the C API.
See PEP 657 for more details. (Contributed by Pablo Galindo, Batuhan Taskaya and Ammar Askar in bpo-43950.)
Note
This feature requires storing column positions in Code Objects,
which may result in a small increase in interpreter memory usage
and disk usage for compiled Python files.
To avoid storing the extra information
and deactivate printing the extra traceback information,
use the -X no_debug_ranges
command line option
or the PYTHONNODEBUGRANGES
environment variable.
PEP 654: Exception Groups and except*
#
PEP 654 introduces language features that enable a program
to raise and handle multiple unrelated exceptions simultaneously.
The builtin types ExceptionGroup
and BaseExceptionGroup
make it possible to group exceptions and raise them together,
and the new except*
syntax generalizes
except
to match subgroups of exception groups.
See PEP 654 for more details.
(Contributed by Irit Katriel in bpo-45292. PEP written by Irit Katriel, Yury Selivanov and Guido van Rossum.)
PEP 678: Exceptions can be enriched with notes#
The add_note()
method is added to BaseException
.
It can be used to enrich exceptions with context information
that is not available at the time when the exception is raised.
The added notes appear in the default traceback.
See PEP 678 for more details.
(Contributed by Irit Katriel in bpo-45607. PEP written by Zac Hatfield-Dodds.)
Other Language Changes#
Starred expressions can be used in for statements. (See bpo-46725 for more details.)
Asynchronous comprehensions are now allowed inside comprehensions in asynchronous functions. Outer comprehensions implicitly become asynchronous. (Contributed by Serhiy Storchaka in bpo-33346.)
A
TypeError
is now raised instead of anAttributeError
incontextlib.ExitStack.enter_context()
andcontextlib.AsyncExitStack.enter_async_context()
for objects which do not support the context manager or asynchronous context manager protocols correspondingly. (Contributed by Serhiy Storchaka in bpo-44471.)A
TypeError
is now raised instead of anAttributeError
inwith
andasync with
statements for objects which do not support the context manager or asynchronous context manager protocols correspondingly. (Contributed by Serhiy Storchaka in bpo-12022.)Added
object.__getstate__()
which provides the default implementation of the__getstate__()
method.Copying
andpickling
instances of subclasses of builtin typesbytearray
,set
,frozenset
,collections.OrderedDict
,collections.deque
,weakref.WeakSet
, anddatetime.tzinfo
now copies and pickles instance attributes implemented as slots. (Contributed by Serhiy Storchaka in bpo-26579.)Add
-P
command line option andPYTHONSAFEPATH
environment variable to not prepend a potentially unsafe path tosys.path
such as the current directory, the script’s directory or an empty string. (Contributed by Victor Stinner in gh-57684.)A
"z"
option was added to the format specification mini-language that coerces negative zero to zero after rounding to the format precision. See PEP 682 for more details. (Contributed by John Belmonte in gh-90153.)Bytes are no longer accepted on
sys.path
. Support broke sometime between Python 3.2 and 3.6 with no one noticing until after Python 3.10.0 was released. Bringing back support would also be problematic due to interactions between-b
andsys.path_importer_cache
when there is a mixture of strings and bytes keys. (Contributed by Thomas Grainger in gh-91181.)
Other CPython Implementation Changes#
Special methods
complex.__complex__()
andbytes.__bytes__()
are implemented to supporttyping.SupportsComplex
andtyping.SupportsBytes
protocols. (Contributed by Mark Dickinson and Dong-hee Na in bpo-24234.)siphash13
is added as a new internal hashing algorithms. It has similar security properties assiphash24
but it is slightly faster for long inputs.str
,bytes
, and some other types now use it as default algorithm forhash()
. PEP 552 hash-based pyc files now usesiphash13
, too. (Contributed by Inada Naoki in bpo-29410.)When an active exception is re-raised by a
raise
statement with no parameters, the traceback attached to this exception is now alwayssys.exc_info()[1].__traceback__
. This means that changes made to the traceback in the currentexcept
clause are reflected in the re-raised exception. (Contributed by Irit Katriel in bpo-45711.)The interpreter state’s representation of handled exceptions (a.k.a exc_info, or _PyErr_StackItem) now has only the
exc_value
field,exc_type
andexc_traceback
have been removed as their values can be derived fromexc_value
. (Contributed by Irit Katriel in bpo-45711.)A new command line option for the Windows installer
AppendPath
has been added. It behaves similiar toPrependPath
but appends the install and scripts directories instead of prepending them. (Contributed by Bastian Neuburger in bpo-44934.)The
PyConfig.module_search_paths_set
field must now be set to 1 for initialization to usePyConfig.module_search_paths
to initializesys.path
. Otherwise, initialization will recalculate the path and replace any values added tomodule_search_paths
.The output of the
--help
option is changed to fit inside 50 lines and 80 columns. Information about Python environment variables and-X options
is available with the new--help-env
or--help-xoptions
flags, and with--help-all
. (Contributed by Éric Araujo in bpo-46142.)
New Modules#
A new module,
tomllib
, was added for parsing TOML. (Contributed by Taneli Hukkinen in bpo-40059.)wsgiref.types
, containing WSGI-specific types for static type checking, was added. (Contributed by Sebastian Rittau in bpo-42012.)
Improved Modules#
asyncio#
Add raw datagram socket functions to the event loop:
sock_sendto()
,sock_recvfrom()
andsock_recvfrom_into()
. (Contributed by Alex Grönholm in bpo-46805.)Add
start_tls()
method for upgrading existing stream-based connections to TLS. (Contributed by Ian Good in bpo-34975.)Add
Barrier
class to the synchronization primitives of the asyncio library. (Contributed by Yves Duprat and Andrew Svetlov in gh-87518.)Add
TaskGroup
class, an asynchronous context manager holding a group of tasks that will wait for all of them upon exit. (Contributed by Yury Seliganov and others.)
contextlib#
Added non parallel-safe chdir()
context manager to change
the current working directory and then restore it on exit. Simple wrapper
around chdir()
. (Contributed by Filipe Laíns in bpo-25625)
dataclasses#
datetime#
Add
datetime.UTC
, a convenience alias fordatetime.timezone.utc
. (Contributed by Kabir Kwatra in gh-91973.)datetime.date.fromisoformat()
,datetime.time.fromisoformat()
anddatetime.datetime.fromisoformat()
can now be used to parse most ISO 8601 formats (barring only those that support fractional hours and minutes). (Contributed by Paul Ganssle in gh-80010.)
enum#
EnumMeta
renamed toEnumType
(EnumMeta
kept as alias).StrEnum
added – enum members are and must be strings.ReprEnum
added – causes only the__repr__
to be modified, not the__str__
nor the__format__
.FlagBoundary
added – controls behavior when invalid values are given to a flag.EnumCheck
added – used byverify
to ensure various constraints.verify
added – function to ensure givenEnumCheck
constraints.member
added – decorator to ensure given object is converted to an enum member.nonmember
added – decorator to ensure given object is not converted to an enum member.property
added – use instead oftypes.DynamicClassAttribute
.global_enum
added – enum decorator to adjust__repr__
and__str__
to show members in the global context – seere.RegexFlag
for an example.Flag
enhancements: members support length, iteration, and containment checks.Enum
/Flag
fixes: members are now defined before__init_subclass__
is called;dir()
now includes methods, etc., from mixed-in data types.Flag
fixes: only primary values (power of two) are considered canonical while composite values (3, 6, 10, etc.) are considered aliases; inverted flags are coerced to their positive equivalent.IntEnum
/IntFlag
/StrEnum
fixes: these now inherit fromReprEnum
so thestr()
output now matchesformat()
output, which is the data types’ (so bothstr(AnIntEnum.ONE)
andformat(AnIntEnum.ONE)
is equal to'1'
).
fractions#
functools#
functools.singledispatch()
now supportstypes.UnionType
andtyping.Union
as annotations to the dispatch argument.:>>> from functools import singledispatch >>> @singledispatch ... def fun(arg, verbose=False): ... if verbose: ... print("Let me just say,", end=" ") ... print(arg) ... >>> @fun.register ... def _(arg: int | float, verbose=False): ... if verbose: ... print("Strength in numbers, eh?", end=" ") ... print(arg) ... >>> from typing import Union >>> @fun.register ... def _(arg: Union[list, set], verbose=False): ... if verbose: ... print("Enumerate this:") ... for i, elem in enumerate(arg): ... print(i, elem) ...
(Contributed by Yurii Karabas in bpo-46014.)
hashlib#
hashlib.blake2b()
andhashlib.blake2s()
now prefer libb2 over Python’s vendored copy. (Contributed by Christian Heimes in bpo-47095.)The internal
_sha3
module with SHA3 and SHAKE algorithms now uses tiny_sha3 instead of the Keccak Code Package to reduce code and binary size. Thehashlib
module prefers optimized SHA3 and SHAKE implementations from OpenSSL. The change affects only installations without OpenSSL support. (Contributed by Christian Heimes in bpo-47098.)Add
hashlib.file_digest()
, a helper function for efficient hashing of files or file-like objects. (Contributed by Christian Heimes in gh-89313.)
IDLE and idlelib#
inspect#
Add
inspect.getmembers_static()
: return all members without triggering dynamic lookup via the descriptor protocol. (Contributed by Weipeng Hong in bpo-30533.)Add
inspect.ismethodwrapper()
for checking if the type of an object is aMethodWrapperType
. (Contributed by Hakan Çelik in bpo-29418.)Change the frame-related functions in the
inspect
module to return a regular object (that is backwards compatible with the old tuple-like interface) that include the extended PEP 657 position information (end line number, column and end column). The affected functions are:inspect.getframeinfo()
,inspect.getouterframes()
,inspect.getinnerframes()
,inspect.stack()
andinspect.trace()
. (Contributed by Pablo Galindo in gh-88116.)
locale#
Add
locale.getencoding()
to get the current locale encoding. It is similar tolocale.getpreferredencoding(False)
but ignores the Python UTF-8 Mode.
math#
Add
math.exp2()
: return 2 raised to the power of x. (Contributed by Gideon Mitchell in bpo-45917.)Add
math.cbrt()
: return the cube root of x. (Contributed by Ajith Ramachandran in bpo-44357.)The behaviour of two
math.pow()
corner cases was changed, for consistency with the IEEE 754 specification. The operationsmath.pow(0.0, -math.inf)
andmath.pow(-0.0, -math.inf)
now returninf
. Previously they raisedValueError
. (Contributed by Mark Dickinson in bpo-44339.)The
math.nan
value is now always available. (Contributed by Victor Stinner in bpo-46917.)
operator#
A new function
operator.call
has been added, such thatoperator.call(obj, *args, **kwargs) == obj(*args, **kwargs)
. (Contributed by Antony Lee in bpo-44019.)
os#
On Windows,
os.urandom()
now usesBCryptGenRandom()
, instead ofCryptGenRandom()
which is deprecated. (Contributed by Dong-hee Na in bpo-44611.)
pathlib#
re#
Atomic grouping (
(?>...)
) and possessive quantifiers (*+
,++
,?+
,{m,n}+
) are now supported in regular expressions. (Contributed by Jeffrey C. Jacobs and Serhiy Storchaka in bpo-433030.)
shutil#
Add optional parameter dir_fd in
shutil.rmtree()
. (Contributed by Serhiy Storchaka in bpo-46245.)
socket#
Add CAN Socket support for NetBSD. (Contributed by Thomas Klausner in bpo-30512.)
create_connection()
has an option to raise, in case of failure to connect, anExceptionGroup
containing all errors instead of only raising the last error. (Contributed by Irit Katriel in bpo-29980.)
sqlite3#
You can now disable the authorizer by passing
None
toset_authorizer()
. (Contributed by Erlend E. Aasland in bpo-44491.)Collation name
create_collation()
can now contain any Unicode character. Collation names with invalid characters now raiseUnicodeEncodeError
instead ofsqlite3.ProgrammingError
. (Contributed by Erlend E. Aasland in bpo-44688.)sqlite3
exceptions now include the SQLite extended error code assqlite_errorcode
and the SQLite error name assqlite_errorname
. (Contributed by Aviv Palivoda, Daniel Shahaf, and Erlend E. Aasland in bpo-16379 and bpo-24139.)Add
setlimit()
andgetlimit()
tosqlite3.Connection
for setting and getting SQLite limits by connection basis. (Contributed by Erlend E. Aasland in bpo-45243.)sqlite3
now setssqlite3.threadsafety
based on the default threading mode the underlying SQLite library has been compiled with. (Contributed by Erlend E. Aasland in bpo-45613.)sqlite3
C callbacks now use unraisable exceptions if callback tracebacks are enabled. Users can now register anunraisable hook handler
to improve their debug experience. (Contributed by Erlend E. Aasland in bpo-45828.)Fetch across rollback no longer raises
InterfaceError
. Instead we leave it to the SQLite library to handle these cases. (Contributed by Erlend E. Aasland in bpo-44092.)Add
serialize()
anddeserialize()
tosqlite3.Connection
for serializing and deserializing databases. (Contributed by Erlend E. Aasland in bpo-41930.)Add
create_window_function()
tosqlite3.Connection
for creating aggregate window functions. (Contributed by Erlend E. Aasland in bpo-34916.)Add
blobopen()
tosqlite3.Connection
.sqlite3.Blob
allows incremental I/O operations on blobs. (Contributed by Aviv Palivoda and Erlend E. Aasland in bpo-24905.)
sys#
sys.exc_info()
now derives thetype
andtraceback
fields from thevalue
(the exception instance), so when an exception is modified while it is being handled, the changes are reflected in the results of subsequent calls toexc_info()
. (Contributed by Irit Katriel in bpo-45711.)Add
sys.exception()
which returns the active exception instance (equivalent tosys.exc_info()[1]
). (Contributed by Irit Katriel in bpo-46328.)Add the
sys.flags.safe_path
flag. (Contributed by Victor Stinner in gh-57684.)
sysconfig#
Three new installation schemes (posix_venv, nt_venv and venv) were added and are used when Python creates new virtual environments or when it is running from a virtual environment. The first two schemes (posix_venv and nt_venv) are OS-specific for non-Windows and Windows, the venv is essentially an alias to one of them according to the OS Python runs on. This is useful for downstream distributors who modify
sysconfig.get_preferred_scheme()
. Third party code that creates new virtual environments should use the new venv installation scheme to determine the paths, as doesvenv
. (Contributed by Miro Hrončok in bpo-45413.)
threading#
On Unix, if the
sem_clockwait()
function is available in the C library (glibc 2.30 and newer), thethreading.Lock.acquire()
method now uses the monotonic clock (time.CLOCK_MONOTONIC
) for the timeout, rather than using the system clock (time.CLOCK_REALTIME
), to not be affected by system clock changes. (Contributed by Victor Stinner in bpo-41710.)
time#
On Unix,
time.sleep()
now uses theclock_nanosleep()
ornanosleep()
function, if available, which has a resolution of 1 nanosecond (10-9 seconds), rather than usingselect()
which has a resolution of 1 microsecond (10-6 seconds). (Contributed by Benjamin Szőke and Victor Stinner in bpo-21302.)On Windows 8.1 and newer,
time.sleep()
now uses a waitable timer based on high-resolution timers which has a resolution of 100 nanoseconds (10-7 seconds). Previously, it had a resolution of 1 millisecond (10-3 seconds). (Contributed by Benjamin Szőke, Dong-hee Na, Eryk Sun and Victor Stinner in bpo-21302 and bpo-45429.)
traceback#
Add
traceback.StackSummary.format_frame_summary()
to allow users to override which frames appear in the traceback, and how they are formatted. (Contributed by Ammar Askar in bpo-44569.)Add
traceback.TracebackException.print()
, which prints the formattedTracebackException
instance to a file. (Contributed by Irit Katriel in bpo-33809.)
typing#
For major changes, see New Features Related to Type Hints.
Add
typing.assert_never()
andtyping.Never
.typing.assert_never()
is useful for asking a type checker to confirm that a line of code is not reachable. At runtime, it raises anAssertionError
. (Contributed by Jelle Zijlstra in gh-90633.)Add
typing.reveal_type()
. This is useful for asking a type checker what type it has inferred for a given expression. At runtime it prints the type of the received value. (Contributed by Jelle Zijlstra in gh-90572.)Add
typing.assert_type()
. This is useful for asking a type checker to confirm that the type it has inferred for a given expression matches the given type. At runtime it simply returns the received value. (Contributed by Jelle Zijlstra in gh-90638.)typing.TypedDict
types can now be generic. (Contributed by Samodya Abeysiriwardane in gh-89026.)NamedTuple
types can now be generic. (Contributed by Serhiy Storchaka in bpo-43923.)Allow subclassing of
typing.Any
. This is useful for avoiding type checker errors related to highly dynamic class, such as mocks. (Contributed by Shantanu Jain in gh-91154.)The
typing.final()
decorator now sets the__final__
attributed on the decorated object. (Contributed by Jelle Zijlstra in gh-90500.)The
typing.get_overloads()
function can be used for introspecting the overloads of a function.typing.clear_overloads()
can be used to clear all registered overloads of a function. (Contributed by Jelle Zijlstra in gh-89263.)The
__init__()
method ofProtocol
subclasses is now preserved. (Contributed by Adrian Garcia Badarasco in gh-88970.)The representation of empty tuple types (
Tuple[()]
) is simplified. This affects introspection, e.g.get_args(Tuple[()])
now evaluates to()
instead of((),)
. (Contributed by Serhiy Storchaka in gh-91137.)Loosen runtime requirements for type annotations by removing the callable check in the private
typing._type_check
function. (Contributed by Gregory Beauregard in gh-90802.)typing.get_type_hints()
now supports evaluating strings as forward references in PEP 585 generic aliases. (Contributed by Niklas Rosenstein in gh-85542.)typing.get_type_hints()
no longer addsOptional
to parameters withNone
as a default. (Contributed by Nikita Sobolev in gh-90353.)typing.get_type_hints()
now supports evaluating bare stringifiedClassVar
annotations. (Contributed by Gregory Beauregard in gh-90711.)typing.no_type_check()
no longer modifies external classes and functions. It also now correctly marks classmethods as not to be type checked. (Contributed by Nikita Sobolev in gh-90729.)
tkinter#
Added method
info_patchlevel()
which returns the exact version of the Tcl library as a named tuple similar tosys.version_info
. (Contributed by Serhiy Storchaka in gh-91827.)
unicodedata#
The Unicode database has been updated to version 14.0.0. (Contributed by Benjamin Peterson in bpo-45190).
unittest#
Added methods
enterContext()
andenterClassContext()
of classTestCase
, methodenterAsyncContext()
of classIsolatedAsyncioTestCase
and functionunittest.enterModuleContext()
. (Contributed by Serhiy Storchaka in bpo-45046.)
venv#
When new Python virtual environments are created, the venv sysconfig installation scheme is used to determine the paths inside the environment. When Python runs in a virtual environment, the same installation scheme is the default. That means that downstream distributors can change the default sysconfig install scheme without changing behavior of virtual environments. Third party code that also creates new virtual environments should do the same. (Contributed by Miro Hrončok in bpo-45413.)
warnings#
warnings.catch_warnings()
now accepts arguments forwarnings.simplefilter()
, providing a more concise way to locally ignore warnings or convert them to errors. (Contributed by Zac Hatfield-Dodds in bpo-47074.)
zipfile#
Added support for specifying member name encoding for reading metadata in the zipfile’s directory and file headers. (Contributed by Stephen J. Turnbull and Serhiy Storchaka in bpo-28080.)
fcntl#
On FreeBSD, the
F_DUP2FD
andF_DUP2FD_CLOEXEC
flags respectively are supported, the former equals todup2
usage while the latter set theFD_CLOEXEC
flag in addition.
Optimizations#
Compiler now optimizes simple C-style formatting with literal format containing only format codes
%s
,%r
and%a
and makes it as fast as corresponding f-string expression. (Contributed by Serhiy Storchaka in bpo-28307.)“Zero-cost” exceptions are implemented. The cost of
try
statements is almost eliminated when no exception is raised. (Contributed by Mark Shannon in bpo-40222.)Pure ASCII strings are now normalized in constant time by
unicodedata.normalize()
. (Contributed by Dong-hee Na in bpo-44987.)math
functionscomb()
andperm()
are now up to 10 times or more faster for large arguments (the speed up is larger for larger k). (Contributed by Serhiy Storchaka in bpo-37295.)Dict don’t store hash value when all inserted keys are Unicode objects. This reduces dict size. For example,
sys.getsizeof(dict.fromkeys("abcdefg"))
becomes 272 bytes from 352 bytes on 64bit platform. (Contributed by Inada Naoki in bpo-46845.)re
’s regular expression matching engine has been partially refactored, and now uses computed gotos (or “threaded code”) on supported platforms. As a result, Python 3.11 executes the pyperformance regular expression benchmarks up to 10% faster than Python 3.10.
Faster CPython#
CPython 3.11 is on average 25% faster than CPython 3.10 when measured with the pyperformance benchmark suite, and compiled with GCC on Ubuntu Linux. Depending on your workload, the speedup could be up to 10-60% faster.
This project focuses on two major areas in Python: faster startup and faster runtime. Other optimizations not under this project are listed in Optimizations.
Faster Startup#
Frozen imports / Static code objects#
Python caches bytecode in the __pycache__ directory to speed up module loading.
Previously in 3.10, Python module execution looked like this:
Read __pycache__ -> Unmarshal -> Heap allocated code object -> Evaluate
In Python 3.11, the core modules essential for Python startup are “frozen”. This means that their code objects (and bytecode) are statically allocated by the interpreter. This reduces the steps in module execution process to this:
Statically allocated code object -> Evaluate
Interpreter startup is now 10-15% faster in Python 3.11. This has a big impact for short-running programs using Python.
(Contributed by Eric Snow, Guido van Rossum and Kumar Aditya in numerous issues.)
Faster Runtime#
Cheaper, lazy Python frames#
Python frames are created whenever Python calls a Python function. This frame holds execution information. The following are new frame optimizations:
Streamlined the frame creation process.
Avoided memory allocation by generously re-using frame space on the C stack.
Streamlined the internal frame struct to contain only essential information. Frames previously held extra debugging and memory management information.
Old-style frame objects are now created only when requested by debuggers or
by Python introspection functions such as sys._getframe
or
inspect.currentframe
. For most user code, no frame objects are
created at all. As a result, nearly all Python functions calls have sped
up significantly. We measured a 3-7% speedup in pyperformance.
(Contributed by Mark Shannon in bpo-44590.)
Inlined Python function calls#
During a Python function call, Python will call an evaluating C function to interpret that function’s code. This effectively limits pure Python recursion to what’s safe for the C stack.
In 3.11, when CPython detects Python code calling another Python function, it sets up a new frame, and “jumps” to the new code inside the new frame. This avoids calling the C interpreting function altogether.
Most Python function calls now consume no C stack space. This speeds up most of such calls. In simple recursive functions like fibonacci or factorial, a 1.7x speedup was observed. This also means recursive functions can recurse significantly deeper (if the user increases the recursion limit). We measured a 1-3% improvement in pyperformance.
(Contributed by Pablo Galindo and Mark Shannon in bpo-45256.)
PEP 659: Specializing Adaptive Interpreter#
PEP 659 is one of the key parts of the faster CPython project. The general idea is that while Python is a dynamic language, most code has regions where objects and types rarely change. This concept is known as type stability.
At runtime, Python will try to look for common patterns and type stability in the executing code. Python will then replace the current operation with a more specialized one. This specialized operation uses fast paths available only to those use cases/types, which generally outperform their generic counterparts. This also brings in another concept called inline caching, where Python caches the results of expensive operations directly in the bytecode.
The specializer will also combine certain common instruction pairs into one superinstruction. This reduces the overhead during execution.
Python will only specialize when it sees code that is “hot” (executed multiple times). This prevents Python from wasting time for run-once code. Python can also de-specialize when code is too dynamic or when the use changes. Specialization is attempted periodically, and specialization attempts are not too expensive. This allows specialization to adapt to new circumstances.
(PEP written by Mark Shannon, with ideas inspired by Stefan Brunthaler. See PEP 659 for more information. Implementation by Mark Shannon and Brandt Bucher, with additional help from Irit Katriel and Dennis Sweeney.)
Operation |
Form |
Specialization |
Operation speedup (up to) |
Contributor(s) |
---|---|---|---|---|
Binary operations |
|
Binary add, multiply and subtract for common types
such as |
10% |
Mark Shannon, Dong-hee Na, Brandt Bucher, Dennis Sweeney |
Subscript |
|
Subscripting container types such as Subscripting custom |
10-25% |
Irit Katriel, Mark Shannon |
Store subscript |
|
Similar to subscripting specialization above. |
10-25% |
Dennis Sweeney |
Calls |
|
Calls to common builtin (C) functions and types such
as |
20% |
Mark Shannon, Ken Jin |
Load global variable |
|
The object’s index in the globals/builtins namespace is cached. Loading globals and builtins require zero namespace lookups. |
Mark Shannon |
|
Load attribute |
|
Similar to loading global variables. The attribute’s index inside the class/object’s namespace is cached. In most cases, attribute loading will require zero namespace lookups. |
Mark Shannon |
|
Load methods for call |
|
The actual address of the method is cached. Method loading now has no namespace lookups – even for classes with long inheritance chains. |
10-20% |
Ken Jin, Mark Shannon |
Store attribute |
|
Similar to load attribute optimization. |
2% in pyperformance |
Mark Shannon |
Unpack Sequence |
|
Specialized for common containers such as |
8% |
Brandt Bucher |
Misc#
Objects now require less memory due to lazily created object namespaces. Their namespace dictionaries now also share keys more freely. (Contributed Mark Shannon in bpo-45340 and bpo-40116.)
A more concise representation of exceptions in the interpreter reduced the time required for catching an exception by about 10%. (Contributed by Irit Katriel in bpo-45711.)
FAQ#
About#
Faster CPython explores optimizations for CPython. The main team is funded by Microsoft to work on this full-time. Pablo Galindo Salgado is also funded by Bloomberg LP to work on the project part-time. Finally, many contributors are volunteers from the community.
CPython bytecode changes#
The bytecode now contains inline cache entries, which take the form of
CACHE
instructions. Many opcodes expect to be followed by an exact number of caches, and instruct the interpreter to skip over them at runtime. Populated caches can look like arbitrary instructions, so great care should be taken when reading or modifying raw, adaptive bytecode containing quickened data.Replaced all numeric
BINARY_*
andINPLACE_*
instructions with a singleBINARY_OP
implementation.Replaced the three call instructions:
CALL_FUNCTION
,CALL_FUNCTION_KW
andCALL_METHOD
withPUSH_NULL
,PRECALL
,CALL
, andKW_NAMES
. This decouples the argument shifting for methods from the handling of keyword arguments and allows better specialization of calls.Removed
COPY_DICT_WITHOUT_KEYS
andGEN_START
.MATCH_CLASS
andMATCH_KEYS
no longer push an additional boolean value indicating whether the match succeeded or failed. Instead, they indicate failure withNone
(where a tuple of extracted values would otherwise be).Replace several stack manipulation instructions (
DUP_TOP
,DUP_TOP_TWO
,ROT_TWO
,ROT_THREE
,ROT_FOUR
, andROT_N
) with newCOPY
andSWAP
instructions.Replaced
JUMP_IF_NOT_EXC_MATCH
byCHECK_EXC_MATCH
which performs the check but does not jump.Replaced
JUMP_IF_NOT_EG_MATCH
byCHECK_EG_MATCH
which performs the check but does not jump.Replaced
JUMP_ABSOLUTE
by the relativeJUMP_BACKWARD
.Added
JUMP_BACKWARD_NO_INTERRUPT
, which is used in certain loops where it is undesirable to handle interrupts.Replaced
POP_JUMP_IF_TRUE
andPOP_JUMP_IF_FALSE
by the relativePOP_JUMP_FORWARD_IF_TRUE
,POP_JUMP_BACKWARD_IF_TRUE
,POP_JUMP_FORWARD_IF_FALSE
andPOP_JUMP_BACKWARD_IF_FALSE
.Added
POP_JUMP_FORWARD_IF_NOT_NONE
,POP_JUMP_BACKWARD_IF_NOT_NONE
,POP_JUMP_FORWARD_IF_NONE
andPOP_JUMP_BACKWARD_IF_NONE
opcodes to speed up conditional jumps.JUMP_IF_TRUE_OR_POP
andJUMP_IF_FALSE_OR_POP
are now relative rather than absolute.RESUME
has been added. It is a no-op. Performs internal tracing, debugging and optimization checks.
Deprecated#
Chaining
classmethod
descriptors (introduced in bpo-19072) is now deprecated. It can no longer be used to wrap other descriptors such asproperty
. The core design of this feature was flawed and caused a number of downstream problems. To “pass-through” aclassmethod
, consider using the__wrapped__
attribute that was added in Python 3.10. (Contributed by Raymond Hettinger in gh-89519.)Octal escapes in string and bytes literals with value larger than
0o377
now produceDeprecationWarning
. In a future Python version they will be aSyntaxWarning
and eventually aSyntaxError
. (Contributed by Serhiy Storchaka in gh-81548.)The
lib2to3
package and2to3
tool are now deprecated and may not be able to parse Python 3.10 or newer. See the PEP 617 (New PEG parser for CPython). (Contributed by Victor Stinner in bpo-40360.)Undocumented modules
sre_compile
,sre_constants
andsre_parse
are now deprecated. (Contributed by Serhiy Storchaka in bpo-47152.)webbrowser.MacOSX
is deprecated and will be removed in Python 3.13. It is untested and undocumented and also not used by webbrowser itself. (Contributed by Dong-hee Na in bpo-42255.)The behavior of returning a value from a
TestCase
andIsolatedAsyncioTestCase
test methods (other than the defaultNone
value), is now deprecated.Deprecated the following
unittest
functions, scheduled for removal in Python 3.13:unittest.findTestCases()
unittest.makeSuite()
unittest.getTestCaseNames()
Use
TestLoader
method instead:(Contributed by Erlend E. Aasland in bpo-5846.)
The
turtle.RawTurtle.settiltangle()
is deprecated since Python 3.1, it now emits a deprecation warning and will be removed in Python 3.13. Useturtle.RawTurtle.tiltangle()
instead (it was earlier incorrectly marked as deprecated, its docstring is now corrected). (Contributed by Hugo van Kemenade in bpo-45837.)The delegation of
int()
to__trunc__()
is now deprecated. Callingint(a)
whentype(a)
implements__trunc__()
but not__int__()
or__index__()
now raises aDeprecationWarning
. (Contributed by Zackery Spytz in bpo-44977.)The following have been deprecated in
configparser
since Python 3.2. Their deprecation warnings have now been updated to note they will removed in Python 3.12:the
configparser.SafeConfigParser
classthe
configparser.ParsingError.filename
propertythe
configparser.RawConfigParser.readfp()
method
(Contributed by Hugo van Kemenade in bpo-45173.)
configparser.LegacyInterpolation
has been deprecated in the docstring since Python 3.2. It now emits aDeprecationWarning
and will be removed in Python 3.13. Useconfigparser.BasicInterpolation
orconfigparser.ExtendedInterpolation
instead. (Contributed by Hugo van Kemenade in bpo-46607.)The
locale.getdefaultlocale()
function is deprecated and will be removed in Python 3.13. Uselocale.setlocale()
,locale.getpreferredencoding(False)
andlocale.getlocale()
functions instead. (Contributed by Victor Stinner in gh-90817.)The
locale.resetlocale()
function is deprecated and will be removed in Python 3.13. Uselocale.setlocale(locale.LC_ALL, "")
instead. (Contributed by Victor Stinner in gh-90817.)The
asynchat
,asyncore
andsmtpd
modules have been deprecated since at least Python 3.6. Their documentation and deprecation warnings have now been updated to note they will removed in Python 3.12 (PEP 594). (Contributed by Hugo van Kemenade in bpo-47022.)PEP 594 led to the deprecations of the following modules which are slated for removal in Python 3.13:
(Contributed by Brett Cannon in bpo-47061 and Victor Stinner in gh-68966.)
More strict rules will be applied now applied for numerical group references and group names in regular expressions in future Python versions. Only sequence of ASCII digits will be now accepted as a numerical reference. The group name in bytes patterns and replacement strings could only contain ASCII letters and digits and underscore. For now, a deprecation warning is raised for such syntax. (Contributed by Serhiy Storchaka in gh-91760.)
typing.Text
, which exists solely to provide compatibility support between Python 2 and Python 3 code, is now deprecated. Its removal is currently unplanned, but users are encouraged to usestr
instead wherever possible. (Contributed by Alex Waygood in gh-92332.)The keyword argument syntax for constructing
TypedDict
types is now deprecated. Support will be removed in Python 3.13. (Contributed by Jingchen Ye in gh-90224.)The
re.template()
function and the correspondingre.TEMPLATE
andre.T
flags are deprecated, as they were undocumented and lacked an obvious purpose. They will be removed in Python 3.13. (Contributed by Serhiy Storchaka and Miro Hrončok in gh-92728.)
Pending Removal in Python 3.12#
The following APIs have been deprecated in earlier Python releases, and will be removed in Python 3.12.
Python API:
PYTHONTHREADDEBUG
importlib.util.set_loader_wrapper()
importlib.util.set_package_wrapper()
importlib.abc.Loader.module_repr()
importlib.abc.Loadermodule_repr()
importlib.machinery.BuiltinImporter.find_module()
importlib.machinery.BuiltinLoader.module_repr()
importlib.machinery.FileFinder.find_module()
importlib.machinery.FrozenImporter.find_module()
importlib.machinery.FrozenLoader.module_repr()
importlib.machinery.WindowsRegistryFinder.find_module()
pathlib.Path.link_to()
The entire distutils namespace
cgi.log()
sqlite3.OptimizedUnicode()
sqlite3.enable_shared_cache()
C API:
PyUnicode_AS_DATA()
PyUnicode_AS_UNICODE()
PyUnicode_AsUnicodeAndSize()
PyUnicode_AsUnicode()
PyUnicode_FromUnicode()
PyUnicode_GET_DATA_SIZE()
PyUnicode_GET_SIZE()
PyUnicode_GetSize()
PyUnicode_IS_COMPACT()
PyUnicode_IS_READY()
Py_UNICODE_WSTR_LENGTH()
_PyUnicode_AsUnicode()
PyUnicode_WCHAR_KIND
PyUnicode_InternImmortal()
Removed#
smtpd.MailmanProxy
is now removed as it is unusable without an external module,mailman
. (Contributed by Dong-hee Na in bpo-35800.)The
binhex
module, deprecated in Python 3.9, is now removed. The followingbinascii
functions, deprecated in Python 3.9, are now also removed:a2b_hqx()
,b2a_hqx()
;rlecode_hqx()
,rledecode_hqx()
.
The
binascii.crc_hqx()
function remains available.(Contributed by Victor Stinner in bpo-45085.)
The distutils
bdist_msi
command, deprecated in Python 3.9, is now removed. Usebdist_wheel
(wheel packages) instead. (Contributed by Hugo van Kemenade in bpo-45124.)Due to significant security concerns, the reuse_address parameter of
asyncio.loop.create_datagram_endpoint()
, disabled in Python 3.9, is now entirely removed. This is because of the behavior of the socket optionSO_REUSEADDR
in UDP. (Contributed by Hugo van Kemenade in bpo-45129.)Removed
__getitem__()
methods ofxml.dom.pulldom.DOMEventStream
,wsgiref.util.FileWrapper
andfileinput.FileInput
, deprecated since Python 3.9. (Contributed by Hugo van Kemenade in bpo-45132.)The following deprecated functions and methods are removed in the
gettext
module:lgettext()
,ldgettext()
,lngettext()
andldngettext()
.Function
bind_textdomain_codeset()
, methodsoutput_charset()
andset_output_charset()
, and the codeset parameter of functionstranslation()
andinstall()
are also removed, since they are only used for thel*gettext()
functions. (Contributed by Dong-hee Na and Serhiy Storchaka in bpo-44235.)The
@asyncio.coroutine
decorator enabling legacy generator-based coroutines to be compatible with async/await code. The function has been deprecated since Python 3.8 and the removal was initially scheduled for Python 3.10. Useasync def
instead. (Contributed by Illia Volochii in bpo-43216.)asyncio.coroutines.CoroWrapper
used for wrapping legacy generator-based coroutine objects in the debug mode. (Contributed by Illia Volochii in bpo-43216.)Removed the deprecated
split()
method of_tkinter.TkappType
. (Contributed by Erlend E. Aasland in bpo-38371.)Removed from the
inspect
module:the
getargspec
function, deprecated since Python 3.0; useinspect.signature()
orinspect.getfullargspec()
instead.the
formatargspec
function, deprecated since Python 3.5; use theinspect.signature()
function andSignature
object directly.the undocumented
Signature.from_builtin
andSignature.from_function
functions, deprecated since Python 3.5; use theSignature.from_callable()
method instead.
(Contributed by Hugo van Kemenade in bpo-45320.)
Remove namespace package support from unittest discovery. It was introduced in Python 3.4 but has been broken since Python 3.7. (Contributed by Inada Naoki in bpo-23882.)
Remove
__class_getitem__
method frompathlib.PurePath
, because it was not used and added by mistake in previous versions. (Contributed by Nikita Sobolev in bpo-46483.)Remove the undocumented private
float.__set_format__()
method, previously known asfloat.__setformat__()
in Python 3.7. Its docstring said: “You probably don’t want to use this function. It exists mainly to be used in Python’s test suite.” (Contributed by Victor Stinner in bpo-46852.)
Porting to Python 3.11#
This section lists previously described changes and other bugfixes that may require changes to your code.
Changes in the Python API#
Prohibited passing non-
concurrent.futures.ThreadPoolExecutor
executors toloop.set_default_executor()
following a deprecation in Python 3.8. (Contributed by Illia Volochii in bpo-43234.)open()
,io.open()
,codecs.open()
andfileinput.FileInput
no longer accept'U'
(“universal newline”) in the file mode. This flag was deprecated since Python 3.3. In Python 3, the “universal newline” is used by default when a file is open in text mode. The newline parameter ofopen()
controls how universal newlines works. (Contributed by Victor Stinner in bpo-37330.)The
pdb
module now reads the.pdbrc
configuration file with the'utf-8'
encoding. (Contributed by Srinivas Reddy Thatiparthy (శ్రీనివాస్ రెడ్డి తాటిపర్తి) in bpo-41137.)calendar
: Thecalendar.LocaleTextCalendar
andcalendar.LocaleHTMLCalendar
classes now uselocale.getlocale()
, instead of usinglocale.getdefaultlocale()
, if no locale is specified. (Contributed by Victor Stinner in bpo-46659.)Global inline flags (e.g.
(?i)
) can now only be used at the start of the regular expressions. Using them not at the start of expression was deprecated since Python 3.6. (Contributed by Serhiy Storchaka in bpo-47066.)re
module: Fix a few long-standing bugs where, in rare cases, capturing group could get wrong result. So the result may be different than before. (Contributed by Ma Lin in bpo-35859.)The population parameter of
random.sample()
must be a sequence. Automatic conversion of sets to lists is no longer supported. If the sample size is larger than the population size, aValueError
is raised. (Contributed by Raymond Hettinger in bpo-40465.)ast.AST
node positions are now validated when provided tocompile()
and other related functions. If invalid positions are detected, aValueError
will be raised. (Contributed by Pablo Galindo in gh-93351)tp_dictoffset
should be treated as write-only. It can be set to describe C extension clases to the VM, but should be regarded as meaningless when read. To get the pointer to the object’s dictionary callPyObject_GenericGetDict()
instead.
Build Changes#
Building Python now requires a C11 compiler. Optional C11 features are not required. (Contributed by Victor Stinner in bpo-46656.)
Building Python now requires support of IEEE 754 floating point numbers. (Contributed by Victor Stinner in bpo-46917.)
CPython can now be built with the ThinLTO option via
--with-lto=thin
. (Contributed by Dong-hee Na and Brett Holman in bpo-44340.)libpython is no longer linked against libcrypt. (Contributed by Mike Gilbert in bpo-45433.)
Building Python now requires a C99
<math.h>
header file providing the following functions:copysign()
,hypot()
,isfinite()
,isinf()
,isnan()
,round()
. (Contributed by Victor Stinner in bpo-45440.)Building Python now requires a C99
<math.h>
header file providing aNAN
constant, or the__builtin_nan()
built-in function. (Contributed by Victor Stinner in bpo-46640.)Building Python now requires support for floating point Not-a-Number (NaN): remove the
Py_NO_NAN
macro. (Contributed by Victor Stinner in bpo-46656.)Freelists for object structs can now be disabled. A new configure option
--without-freelists
can be used to disable all freelists except empty tuple singleton. (Contributed by Christian Heimes in bpo-45522.)Modules/Setup
andModules/makesetup
have been improved and tied up. Extension modules can now be built throughmakesetup
. All except some test modules can be linked statically into main binary or library. (Contributed by Brett Cannon and Christian Heimes in bpo-45548, bpo-45570, bpo-45571, and bpo-43974.)Build dependencies, compiler flags, and linker flags for most stdlib extension modules are now detected by configure. libffi, libnsl, libsqlite3, zlib, bzip2, liblzma, libcrypt, Tcl/Tk, and uuid flags are detected by
pkg-config
(when available).tkinter
now requirespkg-config
command to detect development settings for Tcl/Tk headers and libraries. (Contributed by Christian Heimes and Erlend Egeberg Aasland in bpo-45847, bpo-45747, and bpo-45763.)Note
Use the environment variables
TCLTK_CFLAGS
andTCLTK_LIBS
to manually specify the location of Tcl/Tk headers and libraries. The configure options--with-tcltk-includes
and--with-tcltk-libs
have been removed.On RHEL 7 and CentOS 7 the development packages do not provide
tcl.pc
andtk.pc
, useTCLTK_LIBS="-ltk8.5 -ltkstub8.5 -ltcl8.5"
. The directoryMisc/rhel7
contains.pc
files and instructions how to build Python with RHEL 7’s and CentOS 7’s Tcl/Tk and OpenSSL.CPython now has PEP 11 tier 3 support for cross compiling to WebAssembly platform
wasm32-unknown-emscripten
(Python in the browser). The effort is inspired by previous work like Pyodide. Emscripten provides a limited subset of POSIX APIs. Python standard libraries features and modules related to networking, processes, threading, signals, mmap, and users/groups are not available or don’t work. (Contributed by Christian Heimes and Ethan Smith in gh-84461, promoted in gh-95085)CPython now has PEP 11 tier 3 support for cross compiling to WebAssembly platform
wasm32-unknown-wasi
(WebAssembly System Interface). Like on Emscripten, only a subset of Python’s standard library is available on WASI. (Contributed by Christian Heimes in gh-90473, promoted in gh-95085)CPython will now use 30-bit digits by default for the Python
int
implementation. Previously, the default was to use 30-bit digits on platforms withSIZEOF_VOID_P >= 8
, and 15-bit digits otherwise. It’s still possible to explicitly request use of 15-bit digits via either the--enable-big-digits
option to the configure script or (for Windows) thePYLONG_BITS_IN_DIGIT
variable inPC/pyconfig.h
, but this option may be removed at some point in the future. (Contributed by Mark Dickinson in bpo-45569.)The
tkinter
package now requires Tcl/Tk version 8.5.12 or newer. (Contributed by Serhiy Storchaka in bpo-46996.)
C API Changes#
New Features#
Add a new
PyType_GetName()
function to get type’s short name. (Contributed by Hai Shi in bpo-42035.)Add a new
PyType_GetQualName()
function to get type’s qualified name. (Contributed by Hai Shi in bpo-42035.)Add new
PyThreadState_EnterTracing()
andPyThreadState_LeaveTracing()
functions to the limited C API to suspend and resume tracing and profiling. (Contributed by Victor Stinner in bpo-43760.)Added the
Py_Version
constant which bears the same value asPY_VERSION_HEX
. (Contributed by Gabriele N. Tornetta in bpo-43931.)Py_buffer
and APIs are now part of the limited API and the stable ABI:PyBuffer_CopyData()
bf_getbuffer
andbf_releasebuffer
type slots
(Contributed by Christian Heimes in bpo-45459.)
Added the
PyType_GetModuleByDef
function, used to get the module in which a method was defined, in cases where this information is not available directly (viaPyCMethod
). (Contributed by Petr Viktorin in bpo-46613.)Add new functions to pack and unpack C double (serialize and deserialize):
PyFloat_Pack2()
,PyFloat_Pack4()
,PyFloat_Pack8()
,PyFloat_Unpack2()
,PyFloat_Unpack4()
andPyFloat_Unpack8()
. (Contributed by Victor Stinner in bpo-46906.)Add new functions to get frame object attributes:
PyFrame_GetBuiltins()
,PyFrame_GetGenerator()
,PyFrame_GetGlobals()
,PyFrame_GetLasti()
.Added two new functions to get and set the active exception instance:
PyErr_GetHandledException()
andPyErr_SetHandledException()
. These are alternatives toPyErr_SetExcInfo()
andPyErr_GetExcInfo()
which work with the legacy 3-tuple representation of exceptions. (Contributed by Irit Katriel in bpo-46343.)Added the
PyConfig.safe_path
member. (Contributed by Victor Stinner in gh-57684.)
Porting to Python 3.11#
PyErr_SetExcInfo()
no longer uses thetype
andtraceback
arguments, the interpreter now derives those values from the exception instance (thevalue
argument). The function still steals references of all three arguments. (Contributed by Irit Katriel in bpo-45711.)PyErr_GetExcInfo()
now derives thetype
andtraceback
fields of the result from the exception instance (thevalue
field). (Contributed by Irit Katriel in bpo-45711.)_frozen
has a newis_package
field to indicate whether or not the frozen module is a package. Previously, a negative value in thesize
field was the indicator. Now only non-negative values be used forsize
. (Contributed by Kumar Aditya in bpo-46608.)_PyFrameEvalFunction()
now takes_PyInterpreterFrame*
as its second parameter, instead ofPyFrameObject*
. See PEP 523 for more details of how to use this function pointer type.PyCode_New()
andPyCode_NewWithPosOnlyArgs()
now take an additionalexception_table
argument. Using these functions should be avoided, if at all possible. To get a custom code object: create a code object using the compiler, then get a modified version with thereplace
method.PyCodeObject
no longer has theco_code
,co_varnames
,co_cellvars
andco_freevars
fields. Instead, usePyCode_GetCode()
,PyCode_GetVarnames()
,PyCode_GetCellvars()
andPyCode_GetFreevars()
respectively to access them via the C API. (Contributed by Brandt Bucher in bpo-46841 and Ken Jin in gh-92154 and gh-94936.)The old trashcan macros (
Py_TRASHCAN_SAFE_BEGIN
/Py_TRASHCAN_SAFE_END
) are now deprecated. They should be replaced by the new macrosPy_TRASHCAN_BEGIN
andPy_TRASHCAN_END
.A tp_dealloc function that has the old macros, such as:
static void mytype_dealloc(mytype *p) { PyObject_GC_UnTrack(p); Py_TRASHCAN_SAFE_BEGIN(p); ... Py_TRASHCAN_SAFE_END }
should migrate to the new macros as follows:
static void mytype_dealloc(mytype *p) { PyObject_GC_UnTrack(p); Py_TRASHCAN_BEGIN(p, mytype_dealloc) ... Py_TRASHCAN_END }
Note that
Py_TRASHCAN_BEGIN
has a second argument which should be the deallocation function it is in.To support older Python versions in the same codebase, you can define the following macros and use them throughout the code (credit: these were copied from the
mypy
codebase):#if PY_MAJOR_VERSION >= 3 && PY_MINOR_VERSION >= 8 # define CPy_TRASHCAN_BEGIN(op, dealloc) Py_TRASHCAN_BEGIN(op, dealloc) # define CPy_TRASHCAN_END(op) Py_TRASHCAN_END #else # define CPy_TRASHCAN_BEGIN(op, dealloc) Py_TRASHCAN_SAFE_BEGIN(op) # define CPy_TRASHCAN_END(op) Py_TRASHCAN_SAFE_END(op) #endif
The
PyType_Ready()
function now raises an error if a type is defined with thePy_TPFLAGS_HAVE_GC
flag set but has no traverse function (PyTypeObject.tp_traverse
). (Contributed by Victor Stinner in bpo-44263.)Heap types with the
Py_TPFLAGS_IMMUTABLETYPE
flag can now inherit the PEP 590 vectorcall protocol. Previously, this was only possible for static types. (Contributed by Erlend E. Aasland in bpo-43908)Since
Py_TYPE()
is changed to a inline static function,Py_TYPE(obj) = new_type
must be replaced withPy_SET_TYPE(obj, new_type)
: see thePy_SET_TYPE()
function (available since Python 3.9). For backward compatibility, this macro can be used:#if PY_VERSION_HEX < 0x030900A4 && !defined(Py_SET_TYPE) static inline void _Py_SET_TYPE(PyObject *ob, PyTypeObject *type) { ob->ob_type = type; } #define Py_SET_TYPE(ob, type) _Py_SET_TYPE((PyObject*)(ob), type) #endif
(Contributed by Victor Stinner in bpo-39573.)
Since
Py_SIZE()
is changed to a inline static function,Py_SIZE(obj) = new_size
must be replaced withPy_SET_SIZE(obj, new_size)
: see thePy_SET_SIZE()
function (available since Python 3.9). For backward compatibility, this macro can be used:#if PY_VERSION_HEX < 0x030900A4 && !defined(Py_SET_SIZE) static inline void _Py_SET_SIZE(PyVarObject *ob, Py_ssize_t size) { ob->ob_size = size; } #define Py_SET_SIZE(ob, size) _Py_SET_SIZE((PyVarObject*)(ob), size) #endif
(Contributed by Victor Stinner in bpo-39573.)
<Python.h>
no longer includes the header files<stdlib.h>
,<stdio.h>
,<errno.h>
and<string.h>
when thePy_LIMITED_API
macro is set to0x030b0000
(Python 3.11) or higher. C extensions should explicitly include the header files after#include <Python.h>
. (Contributed by Victor Stinner in bpo-45434.)The non-limited API files
cellobject.h
,classobject.h
,code.h
,context.h
,funcobject.h
,genobject.h
andlongintrepr.h
have been moved to theInclude/cpython
directory. Moreover, theeval.h
header file was removed. These files must not be included directly, as they are already included inPython.h
: Include Files. If they have been included directly, consider includingPython.h
instead. (Contributed by Victor Stinner in bpo-35134.)The
PyUnicode_CHECK_INTERNED()
macro has been excluded from the limited C API. It was never usable there, because it used internal structures which are not available in the limited C API. (Contributed by Victor Stinner in bpo-46007.)The following frame functions and type are now directly available with
#include <Python.h>
, it’s no longer needed to add#include <frameobject.h>
:PyFrame_Check()
PyFrame_Type
(Contributed by Victor Stinner in gh-93937.)
The
PyFrameObject
structure members have been removed from the public C API.While the documentation notes that the
PyFrameObject
fields are subject to change at any time, they have been stable for a long time and were used in several popular extensions.In Python 3.11, the frame struct was reorganized to allow performance optimizations. Some fields were removed entirely, as they were details of the old implementation.
PyFrameObject
fields:f_back
: usePyFrame_GetBack()
.f_blockstack
: removed.f_builtins
: usePyFrame_GetBuiltins()
.f_code
: usePyFrame_GetCode()
.f_gen
: usePyFrame_GetGenerator()
.f_globals
: usePyFrame_GetGlobals()
.f_iblock
: removed.f_lasti
: usePyFrame_GetLasti()
. Code usingf_lasti
withPyCode_Addr2Line()
should usePyFrame_GetLineNumber()
instead; it may be faster.f_lineno
: usePyFrame_GetLineNumber()
f_locals
: usePyFrame_GetLocals()
.f_stackdepth
: removed.f_state
: no public API (renamed tof_frame.f_state
).f_trace
: no public API.f_trace_lines
: usePyObject_GetAttrString((PyObject*)frame, "f_trace_lines")
.f_trace_opcodes
: usePyObject_GetAttrString((PyObject*)frame, "f_trace_opcodes")
.f_localsplus
: no public API (renamed tof_frame.localsplus
).f_valuestack
: removed.
The Python frame object is now created lazily. A side effect is that the
f_back
member must not be accessed directly, since its value is now also computed lazily. ThePyFrame_GetBack()
function must be called instead.Debuggers that accessed the
f_locals
directly must callPyFrame_GetLocals()
instead. They no longer need to callPyFrame_FastToLocalsWithError()
orPyFrame_LocalsToFast()
, in fact they should not call those functions. The necessary updating of the frame is now managed by the virtual machine.Code defining
PyFrame_GetCode()
on Python 3.8 and older:#if PY_VERSION_HEX < 0x030900B1 static inline PyCodeObject* PyFrame_GetCode(PyFrameObject *frame) { Py_INCREF(frame->f_code); return frame->f_code; } #endif
Code defining
PyFrame_GetBack()
on Python 3.8 and older:#if PY_VERSION_HEX < 0x030900B1 static inline PyFrameObject* PyFrame_GetBack(PyFrameObject *frame) { Py_XINCREF(frame->f_back); return frame->f_back; } #endif
Or use the pythoncapi_compat project to get these two functions on older Python versions.
Changes of the
PyThreadState
structure members:frame
: removed, usePyThreadState_GetFrame()
(function added to Python 3.9 by bpo-40429). Warning: the function returns a strong reference, need to callPy_XDECREF()
.tracing
: changed, usePyThreadState_EnterTracing()
andPyThreadState_LeaveTracing()
(functions added to Python 3.11 by bpo-43760).recursion_depth
: removed, use(tstate->recursion_limit - tstate->recursion_remaining)
instead.stackcheck_counter
: removed.
Code defining
PyThreadState_GetFrame()
on Python 3.8 and older:#if PY_VERSION_HEX < 0x030900B1 static inline PyFrameObject* PyThreadState_GetFrame(PyThreadState *tstate) { Py_XINCREF(tstate->frame); return tstate->frame; } #endif
Code defining
PyThreadState_EnterTracing()
andPyThreadState_LeaveTracing()
on Python 3.10 and older:#if PY_VERSION_HEX < 0x030B00A2 static inline void PyThreadState_EnterTracing(PyThreadState *tstate) { tstate->tracing++; #if PY_VERSION_HEX >= 0x030A00A1 tstate->cframe->use_tracing = 0; #else tstate->use_tracing = 0; #endif } static inline void PyThreadState_LeaveTracing(PyThreadState *tstate) { int use_tracing = (tstate->c_tracefunc != NULL || tstate->c_profilefunc != NULL); tstate->tracing--; #if PY_VERSION_HEX >= 0x030A00A1 tstate->cframe->use_tracing = use_tracing; #else tstate->use_tracing = use_tracing; #endif } #endif
Or use the pythoncapi_compat project to get these functions on old Python functions.
Distributors are encouraged to build Python with the optimized Blake2 library libb2.
The
PyConfig.module_search_paths_set
field must now be set to 1 for initialization to usePyConfig.module_search_paths
to initializesys.path
. Otherwise, initialization will recalculate the path and replace any values added tomodule_search_paths
.PyConfig_Read()
no longer calculates the initial search path, and will not fill any values intoPyConfig.module_search_paths
. To calculate default paths and then modify them, finish initialization and usePySys_GetObject()
to retrievesys.path
as a Python list object and modify it directly.
Deprecated#
Deprecate the following functions to configure the Python initialization:
PySys_HasWarnOptions()
_Py_SetProgramFullPath()
Use the new
PyConfig
API of the Python Initialization Configuration instead (PEP 587). (Contributed by Victor Stinner in gh-88279.)Deprecate the
ob_shash
member of thePyBytesObject
. UsePyObject_Hash()
instead. (Contributed by Inada Naoki in bpo-46864.)
Removed#
PyFrame_BlockSetup()
andPyFrame_BlockPop()
have been removed. (Contributed by Mark Shannon in bpo-40222.)Remove the following math macros using the
errno
variable:Py_ADJUST_ERANGE1()
Py_ADJUST_ERANGE2()
Py_OVERFLOWED()
Py_SET_ERANGE_IF_OVERFLOW()
Py_SET_ERRNO_ON_MATH_ERROR()
(Contributed by Victor Stinner in bpo-45412.)
Remove
Py_UNICODE_COPY()
andPy_UNICODE_FILL()
macros, deprecated since Python 3.3. UsePyUnicode_CopyCharacters()
ormemcpy()
(wchar_t*
string), andPyUnicode_Fill()
functions instead. (Contributed by Victor Stinner in bpo-41123.)Remove the
pystrhex.h
header file. It only contains private functions. C extensions should only include the main<Python.h>
header file. (Contributed by Victor Stinner in bpo-45434.)Remove the
Py_FORCE_DOUBLE()
macro. It was used by thePy_IS_INFINITY()
macro. (Contributed by Victor Stinner in bpo-45440.)The following items are no longer available when
Py_LIMITED_API
is defined:the
Py_MARSHAL_VERSION
macro
These are not part of the limited API.
(Contributed by Victor Stinner in bpo-45474.)
Exclude
PyWeakref_GET_OBJECT()
from the limited C API. It never worked since thePyWeakReference
structure is opaque in the limited C API. (Contributed by Victor Stinner in bpo-35134.)Remove the
PyHeapType_GET_MEMBERS()
macro. It was exposed in the public C API by mistake, it must only be used by Python internally. Use thePyTypeObject.tp_members
member instead. (Contributed by Victor Stinner in bpo-40170.)Remove the
HAVE_PY_SET_53BIT_PRECISION
macro (moved to the internal C API). (Contributed by Victor Stinner in bpo-45412.)Remove the
Py_UNICODE
encoder APIs, as they have been deprecated since Python 3.3, are little used and are inefficient relative to the recommended alternatives.The removed functions are:
PyUnicode_Encode()
PyUnicode_EncodeASCII()
PyUnicode_EncodeLatin1()
PyUnicode_EncodeUTF7()
PyUnicode_EncodeUTF8()
PyUnicode_EncodeUTF16()
PyUnicode_EncodeUTF32()
PyUnicode_EncodeUnicodeEscape()
PyUnicode_EncodeRawUnicodeEscape()
PyUnicode_EncodeCharmap()
PyUnicode_TranslateCharmap()
PyUnicode_EncodeDecimal()
PyUnicode_TransformDecimalToASCII()
See PEP 624 for details and migration guidance. (Contributed by Inada Naoki in bpo-44029.)