跳转到主要内容

What's new in Python 3.16

Editor
Savannah Ostrowski

* Anyone can add text to this document. Do not spend very much time on the wording of your changes, because your text will probably get rewritten to some degree.

* The maintainer will go through Misc/NEWS periodically and add changes; it's therefore more important to add your changes to Misc/NEWS than to this file.

* This is not a complete list of every single change; completeness is the purpose of Misc/NEWS. Some changes I consider too small or esoteric to include. If such a change is added to the text, I'll just remove it. (This is another reason you shouldn't spend too much time on writing your addition.)

* If you want to draw your new text to the attention of the maintainer, add 'XXX' to the beginning of the paragraph or section.

* It's OK to just add a fragmentary note about a change. For example: "XXX Describe the transmogrify() function added to the socket module." The maintainer will research the change and write the necessary text.

* You can comment out your additions if you like, but it's not necessary (especially when a final release is some months away).

* Credit the author of a patch or bugfix. Just the name is sufficient; the e-mail address isn't necessary.

  • It's helpful to add the issue number as a comment:

XXX Describe the transmogrify() function added to the socket module. (Contributed by P.Y. Developer in 12345.)

This saves the maintainer the effort of going through the VCS log when researching a change.

This article explains the new features in Python 3.16, compared to 3.15.

For full details, see the changelog <changelog>.

[!NOTE] Prerelease users should be aware that this document is currently in draft form. It will be updated substantially as Python 3.16 moves towards release, so it's worth checking back even after reading earlier versions.

Summary -- Release highlights

New features

Other language changes

  • The iter function now accepts the stop_exception parameter. The created iterator stops when the callable raises the specified exception. The second parameter is now named stop_value and can be passed by keyword. aiter now accepts the same stop_value and stop_exception parameters, calling an asynchronous callable and awaiting the result. (Contributed by Serhiy Storchaka in 64862.)
  • memoryview.cast now allows casting a multidimensional F-contiguous view to a one-dimensional view. (Contributed by Jaemin Park in 91484.)
  • memoryview.cast now accepts an order parameter. With order='F' it returns a zero-copy Fortran-contiguous (column-major) view of a flat buffer, which is useful for interfacing with column-major libraries. (Contributed by Serhiy Storchaka in 78959.)
  • Frame objects <frame-objects> now support weak references <weakref>. This allows associating extra data with active frames, for example in debuggers, without keeping the frames (and everything they reference) alive indefinitely. (Contributed by Łukasz Langa in 102960.)
  • TypeError messages for some instance methods called with too many positional arguments now suggest checking whether the method definition is missing the self parameter. (Contributed by Suren Nihalani in 152315.)
  • anext and aiter are now implemented in Python instead of C. The awaitable returned by anext(it, default) and the __anext__() of the iterator returned by aiter(callable, ...) are now plain coroutines, so introspection tools such as asyncio.print_call_graph can see through them into the awaited code. (Contributed by Kumar Aditya in 157361.)

Default interactive shell

  • The REPL now displays error messages when editing commands cannot be performed, such as when tab completion finds no matches or a history search fails. (Contributed by Bartosz Sławecki in 148228.)

New modules

  • None yet.

Improved modules

argparse

  • ~argparse.ArgumentGroup and ~argparse.MutuallyExclusiveGroup, the types returned by ~argparse.ArgumentParser.add_argument_group and ~argparse.ArgumentParser.add_mutually_exclusive_group, are now public. The _-prefixed names remain as aliases. (Contributed by Bernát Gábor in 144812.)

asyncio

  • Add the mode parameter to asyncio.loop.create_unix_server and asyncio.start_unix_server to set the permissions of the Unix socket file created for path. (Contributed by Sam Bull in 94984.)
  • Add eager_tasks parameter to asyncio.run function and asyncio.Runner class constructor. (Contributed by Andrew Svetlov in 128289.)

bdb

  • bdb.BdbQuit is now a subclass of BaseException, instead of Exception, to allow it to be used for cleanly exiting the debugger without being accidentally caught by user code. (Contributed by Tian Gao in 149337.)

bz2

  • Added constants ~bz2.bzlib_version and ~bz2.bzlib_version_info which provide information about the version of the bzip2 compression library in use. (Contributed by Serhiy Storchaka in 117404.)

codecs

  • On platforms that provide the C library's iconv(3) function, every encoding known to iconv for which Python has no built-in codec is now available (for example cp1133). Prefixing an encoding name with iconv: forces the iconv-based codec even when a built-in codec of the same name exists. (Contributed by Serhiy Storchaka in 152997.)

compression.zstd

  • Added constants ~compression.zstd.ZSTD_VERSION and ~compression.zstd.ZSTD_VERSION_INFO which provide information about the version of the zstd library that was used for building the module. ~compression.zstd.zstd_version_info is now a named tuple. (Contributed by Serhiy Storchaka in 117404.)

concurrent.futures

  • The iterator returned by concurrent.futures.Executor.map is no longer automatically closed if a function call raises an exception. Use the !close method to explicitly close the iterator. (Contributed by xzmeng and Serhiy Storchaka in 108518.)

concurrent.interpreters

  • ~concurrent.interpreters.Interpreter objects now support the context manager protocol. Exiting the with block calls ~concurrent.interpreters.Interpreter.close. (Contributed by Peter Bierma in 158370.)

csv

  • csv.Sniffer.sniff now deduces the dialect by trial parsing instead of heuristics based on matching isolated fragments and on character frequencies, so the result is consistent with how csv.reader will parse the sample. It can now detect the escapechar parameter, accepts non-ASCII delimiters in the delimiters argument, no longer misdetects the delimiter when the sample contains delimiter characters inside quoted fields, and no longer takes quadratic time on quoted samples. The results may differ from those of earlier Python versions. (Contributed by Serhiy Storchaka in 83273.)
  • csv.Sniffer.sniff now detects the lineterminator parameter by a majority vote among the line endings of the sample, instead of always returning '\r\n'. (Contributed by Serhiy Storchaka in 75008.)

ctypes

  • Add ctypes.util.struct for generating ~ctypes.Structure types from an annotation-based syntax, similar to how the dataclasses module is used. (Contributed by Peter Bierma in 104533.)
  • Add ctypes.util.wrap_dll_function for generating function pointers through a function signature. (Contributed by Peter Bierma in 153903.)
  • Added constants ~ctypes.LIBFFI_VERSION, ~ctypes.libffi_version, ~ctypes.LIBFFI_VERSION_INFO and ~ctypes.libffi_version_info, which provide information about the version of the libffi library in use. They are only available with libffi 3.5 or later. (Contributed by Serhiy Storchaka in 157487.)
  • Objects with an !_as_parameter_ attribute are now implicitly converted when passed for a ~ctypes.POINTER argument. (Contributed by Peter Bierma in 71990.)
  • Callback functions <ctypes-callback-functions> can now return ~ctypes.Structure and ~ctypes.Union instances by value. (Contributed by Russell Keith-Magee in 49960.)

curses

  • The curses character-cell window methods now accept a full character cell --- a spacing character optionally followed by combining characters --- in addition to a single integer or byte character. This affects ~curses.window.addch, ~curses.window.bkgd, ~curses.window.bkgdset, ~curses.window.border, ~curses.window.box, ~curses.window.echochar, ~curses.window.hline, ~curses.window.insch and ~curses.window.vline. Also add the wide-character read methods ~curses.window.get_wstr and ~curses.window.in_wstr, the counterparts of ~curses.window.getstr and ~curses.window.instr that return a str rather than bytes, and the module functions curses.erasewchar, curses.killwchar and curses.wunctrl, the wide-character counterparts of curses.erasechar, curses.killchar and curses.unctrl. On a narrow (non-ncursesw) build the character cell holds a single character without combining marks, representable as one byte in the window's encoding, and ~curses.window.in_wstr returns its decoded text. (Contributed by Serhiy Storchaka in 151757.)
  • Add the curses.complexchar type, representing a styled character cell (its text, attributes and color pair), and the window methods ~curses.window.in_wch and ~curses.window.getbkgrnd that return one --- the counterparts of ~curses.window.inch and ~curses.window.getbkgd. The character-cell methods, such as ~curses.window.addch and ~curses.window.border, now also accept a ~curses.complexchar. These work whether or not Python was built against a wide-character-aware curses library; on a narrow build a cell holds a single character representable as one byte in the window's encoding (so only 8-bit locales are supported). (Contributed by Serhiy Storchaka in 152233.)
  • Add the curses.complexstr type, an immutable run of styled cells (the string counterpart of ~curses.complexchar), and the window method ~curses.window.in_wchstr that returns one. The string-cell methods ~curses.window.addstr, ~curses.window.addnstr, ~curses.window.insstr and ~curses.window.insnstr now also accept a ~curses.complexstr. Like ~curses.complexchar, it works whether or not Python was built against a wide-character-aware curses library. (Contributed by Serhiy Storchaka in 152233.)
  • The wide-character curses functions and methods ~curses.window.get_wch, ~curses.window.get_wstr, curses.unget_wch, curses.erasewchar, curses.killwchar and curses.wunctrl now also work when Python is not built against a wide-character-aware curses library, on an 8-bit locale, where each character is a single byte in the relevant encoding. curses.ungetch now also accepts a one-character string, like curses.unget_wch; on a wide-character build it can be any character (previously a multibyte character raised OverflowError). (Contributed by Serhiy Storchaka in 152470.)
  • Add support for multiple terminals to the curses module: the new functions curses.newterm, curses.set_term and curses.new_prescr, the corresponding screen <curses-screen-objects> object, and the window.use() <curses.window.use> method. (Contributed by Serhiy Storchaka in 90092.)
  • Add the curses window methods ~curses.window.attr_get, ~curses.window.attr_set, ~curses.window.attr_on, ~curses.window.attr_off and ~curses.window.color_set, which pass the color pair as a separate argument instead of packing it into the attribute value, and the corresponding WA_* attribute constants. (Contributed by Serhiy Storchaka in 152219.)
  • Add the curses.term_attrs function, which returns the supported video attributes as WA_* <curses-wa-constants> values, the counterpart of curses.termattrs. (Contributed by Serhiy Storchaka in 152332.)
  • Add the WACS_* constants to the curses module, the counterparts of the ACS_* <curses-acs-codes> line-drawing codes as curses.complexchar cells. (Contributed by Serhiy Storchaka in 155863.)
  • Add the curses functions curses.alloc_pair, curses.find_pair, curses.free_pair and curses.reset_color_pairs for dynamic color-pair management, available when built against a wide-character ncurses with extended-color support. (Contributed by Serhiy Storchaka in 151774.)
  • Add the curses functions ~curses.scr_dump, ~curses.scr_restore, ~curses.scr_init and ~curses.scr_set, which dump the whole screen to a file and restore it. (Contributed by Serhiy Storchaka in 152260.)
  • Add the curses window method ~curses.window.dupwin, which returns a new window that is an independent duplicate of an existing one. (Contributed by Serhiy Storchaka in 152258.)
  • Add the soft-label-key functions to the curses module, which manage a row of labels along the bottom line of the screen: ~curses.slk_init, ~curses.slk_set, ~curses.slk_label, ~curses.slk_refresh, ~curses.slk_noutrefresh, ~curses.slk_clear, ~curses.slk_restore, ~curses.slk_touch, the attribute functions ~curses.slk_attron, ~curses.slk_attroff, ~curses.slk_attrset, ~curses.slk_attr, ~curses.slk_attr_on, ~curses.slk_attr_off, ~curses.slk_attr_set, and ~curses.slk_color. (Contributed by Serhiy Storchaka in 152263.)
  • Add the curses key-management functions ~curses.define_key, ~curses.key_defined and ~curses.keyok, available when built against an ncurses with NCURSES_EXT_FUNCS. (Contributed by Serhiy Storchaka in 152334.)
  • Add the curses.has_mouse function and the curses.window.mouse_trafo method, completing the curses mouse interface. (Contributed by Serhiy Storchaka in 152325.)
  • Add curses functions and window methods that report state which could previously only be set, such as curses.window.is_keypad, curses.window.getparent and curses.is_cbreak, available when built against an ncurses with NCURSES_EXT_FUNCS. (Contributed by Serhiy Storchaka in 151776.)
  • Add curses.nofilter, which undoes the effect of curses.filter. (Contributed by Serhiy Storchaka in 151744.)
  • curses.textpad.Textbox now supports entering and reading back the full Unicode range, including combining characters, when curses is built with wide-character support. (Contributed by Serhiy Storchaka in 133031.)

datetime

  • Add the %Nf format code (where N is from 1 to 6) to datetime.datetime.strftime, datetime.time.strftime, datetime.date.strftime, datetime.datetime.strptime and datetime.time.strptime. (Contributed by Stan Ulbrych and Karoline Pauls in 157290.)
  • Add datetime.timezone.name, the name passed to the constructor, or None if no name was passed. (Contributed by Stan Ulbrych in 131210.)

dbm.gnu

  • Added constants ~dbm.gnu.GDBM_VERSION_INFO, ~dbm.gnu.gdbm_version_info and ~dbm.gnu.gdbm_version, which provide information about the version of the GDBM library in use. (Contributed by Serhiy Storchaka in 157478.)

dbm.ndbm

  • Added constants which provide information about the version of the underlying library in use: ~dbm.ndbm.GDBM_VERSION_INFO, ~dbm.ndbm.gdbm_version_info and ~dbm.ndbm.gdbm_version if it is GDBM, or ~dbm.ndbm.BDB_VERSION, ~dbm.ndbm.bdb_version, ~dbm.ndbm.BDB_VERSION_INFO and ~dbm.ndbm.bdb_version_info if it is Berkeley DB. dbm.ndbm.library is now 'ndbm' instead of 'GNU gdbm' for a classic NDBM library. (Contributed by Serhiy Storchaka in 157480.)

decimal

  • Added constants ~decimal.LIBMPDEC_VERSION, ~decimal.libmpdec_version, ~decimal.LIBMPDEC_VERSION_INFO, and ~decimal.libmpdec_version_info in the C implementation, which provide information about the version of the libmpdec library in use. (Contributed by Serhiy Storchaka in 157475.)

difflib

  • Expose the optional autojunk parameter of difflib.SequenceMatcher in public functions and class methods in difflib, allowing callers to control the automatic junk heuristic. (Contributed by Tomasz Kazimierczak in 118150.)

encodings

  • Add the utf-7-imap codec, implementing the modified UTF-7 encoding used for international IMAP4 mailbox names (3501). (Contributed by Serhiy Storchaka in 66788.)
  • encodings.idna.nameprep now takes a limit argument that allows rejecting extremely large input early. (Contributed by Petr Viktorin in 157675.)

gzip

  • gzip.open now accepts an optional argument mtime which is passed on to the constructor of the ~gzip.GzipFile class. (Contributed by Marin Misur in 91372.)

imaplib

  • Add the ~imaplib.IMAP4.id method, a wrapper for the ID command (2971). (Contributed by Serhiy Storchaka in 98092.)
  • Add the ~imaplib.IMAP4.move method, a wrapper for the MOVE command (6851). (Contributed by Serhiy Storchaka in 77508.)
  • Add the ~imaplib.IMAP4.login_plain method, which authenticates using the PLAIN SASL mechanism (4616) and supports non-ASCII user names and passwords. (Contributed by Przemysław Buczkowski and Serhiy Storchaka in 89869.)
  • Non-ASCII mailbox names are now automatically encoded as modified UTF-7 (3501, section 5.1.3) in the default mode, so international mailbox names can be passed as ordinary str without enabling UTF8=ACCEPT (under which they are sent as UTF-8). (Contributed by Serhiy Storchaka in 49555.)
  • The ~imaplib.IMAP4.copy, ~imaplib.IMAP4.move, ~imaplib.IMAP4.fetch, ~imaplib.IMAP4.store, ~imaplib.IMAP4.search, ~imaplib.IMAP4.sort, ~imaplib.IMAP4.thread and ~imaplib.IMAP4.expunge methods now accept a keyword-only uid argument that selects the corresponding UID command, as a more convenient alternative to ~imaplib.IMAP4.uid. (Contributed by Serhiy Storchaka in 153502.)
  • ~imaplib.IMAP4.search, ~imaplib.IMAP4.sort and ~imaplib.IMAP4.thread (and the corresponding uid commands) now encode str search criteria to the declared charset, so international search text can be passed as an ordinary str. When charset is None (as it must be under UTF8=ACCEPT), the criteria are sent using the connection encoding instead. (Contributed by Serhiy Storchaka in 153494.)
  • Command methods now accept structured arguments, so the module takes care of quoting instead of the caller. A message_set and lists of flags or other atoms can be passed as sequences instead of preformatted strings, and the ~imaplib.IMAP4.search, ~imaplib.IMAP4.fetch, ~imaplib.IMAP4.sort, ~imaplib.IMAP4.thread and ~imaplib.IMAP4.uid methods accept a params keyword argument that substitutes and quotes ? placeholders, in the manner of sqlite3 parameter substitution. (Contributed by Serhiy Storchaka in 153521.)

io

  • Add io.BytesIO.peek method to read without advancing position. (Contributed by Marcel Martin in 90533.)

ipaddress

  • Add IPv4Network.next_network() <ipaddress.IPv4Network.next_network> and IPv6Network.next_network() <ipaddress.IPv6Network.next_network> methods to find the next nearest network with a specific prefix size. (Contributed by Faisal Mahmood in 87027.)

logging

  • ~logging.handlers.TimedRotatingFileHandler now uses the creation time instead of the last modification time of an existing log file as the basis for the first rotation after handler creation, if supported by the OS and file system. This allows it to be used in short-running programs that start and end before the rotation interval expires. (Contributed by Iván Márton and Serhiy Storchaka in 84649.)

lzma

  • Add support of new BCJ filters ARM64 and RISC-V via !lzma.FILTER_ARM64 and !lzma.FILTER_RISCV. Note that the new filters will work only if the runtime library supports them. ARM64 filter requires lzma 5.4.0 or newer while RISC-V requires 5.6.0 or newer. (Contributed by Chien Wong in 115988.)
  • Added constants ~lzma.LZMA_VERSION, ~lzma.lzma_version, ~lzma.LZMA_VERSION_INFO, and ~lzma.lzma_version_info, which provide information about the version of the lzma library in use. (Contributed by Serhiy Storchaka in 117404.)

math

  • Added trigonometric functions that work in units of half turns, rather than radians. The new functions math.acospi, math.asinpi, math.atanpi, and math.atan2pi return half-turn angles. The new functions math.cospi, math.sinpi, and math.tanpi take half-turn angle arguments. These functions are recommended by IEEE 754-2019 and standardized in C23. (Contributed by Jeff Epler in 150534.)

multiprocessing

  • Add the optional buffersize parameter to multiprocessing.pool.Pool.imap and multiprocessing.pool.Pool.imap_unordered to limit the number of submitted tasks whose results have not yet been yielded. If the buffer is full, iteration over the iterable pauses until a result is yielded from the buffer. To fully utilize pool's capacity when using this feature, set buffersize at least to the number of processes in pool (to consume iterable as you go), or even higher (to prefetch the next N=buffersize-processes arguments). (Contributed by Oleksandr Baltian in 136871.)

os

  • Add os.pidfd_getfd for duplicating a file descriptor from another process via a pidfd. Available on Linux 5.6+. (Contributed by Maurycy Pawłowski-Wieroński in 149464.)
  • The following functions are now available on Android: os.chroot, os.initgroups, os.setegid, os.seteuid, os.setgid, os.setregid, os.setresgid, os.setresuid, os.setreuid, and os.setuid. Calling them without sufficient privileges now raises PermissionError instead of the functions being unavailable. (Contributed by Md Arif in 152936.)
  • os.readlink now accepts a file descriptor referring to a symlink on Linux, Android, macOS and iOS. (Contributed by OOTS in 157899.)

pydoc

  • Modernize the HTML output of pydoc. The pages generated by the pydoc HTTP server now use semantic HTML5 markup and a new style sheet based on the python-docs-theme used by docs.python.org, with dark mode support. Class members inherited from other classes are collapsed by default. (Contributed by Serhiy Storchaka in 153906.)

pickle

  • Weak references (weakref.ref), weakref.WeakMethod, weakref.WeakSet, weakref.WeakValueDictionary and weakref.WeakKeyDictionary can now be pickled if the weakly referenced objects are strongly referenced elsewhere in the same pickle; otherwise pickling raises pickle.PicklingError. !WeakMethod and !WeakSet can now also be copied. (Contributed by Serhiy Storchaka in 74876.)

quopri

  • quopri.decode, quopri.decodestring and binascii.a2b_qp gained a strip_ws parameter. When true, trailing whitespace is stripped from each line while decoding, as required by 2045 for a quoted-printable body. (Contributed by Serhiy Storchaka in 62222.)

re

  • re now supports set operations and nested sets in character classes, as described in Unicode Technical Standard #18: set difference ([A--B]), intersection ([A&&B]) and union ([A||B]), where an operand may be a nested set written in square brackets. For example, [a-z--[aeiou]] matches an ASCII lowercase consonant. (Contributed by Serhiy Storchaka in 152100.)
  • Regular expressions now support Unicode property escapes \p{...} and \P{...}, which match a character by a Unicode property -- for example \p{Lu} (an uppercase letter), \p{Cased} or \p{ASCII}. See the regular expression syntax <re-syntax> for the supported properties. (Contributed by Serhiy Storchaka in 95555.)

readline

  • Added constants ~readline.READLINE_VERSION_INFO, ~readline.readline_version_info and ~readline.readline_version, which provide information about the version of the Readline library in use. (Contributed by Serhiy Storchaka in 157485.)

shlex

  • Add keyword-only parameter force to shlex.quote to force quoting a string, even if it is already safe for a shell without being quoted. (Contributed by Jay Berry in 148846.)

shutil

  • Add keyword-only copy_dir_stat parameter to shutil.copytree: set it to false to skip copying directory metadata with shutil.copystat, e.g. when copying a read-only tree. (Contributed by Łukasz Langa in 89721.)
  • shutil.unpack_archive can now unpack wheel files using the whl format. (Contributed by Emma Harper Smith and Stan Ulbrych in 84027.)

socket

  • The socket.sethostname function is now available on Android. Calling it without sufficient privileges now raises PermissionError instead of the function being unavailable. (Contributed by Md Arif in 152936.)

sqlite3

  • sqlite3.Blob now supports negative-step slices for reading and writing (for example, blob[9:0:-2]). Previously, such slices would raise SystemError or ValueError. (Contributed by Jiseok CHOI in 150449.)
  • Added constants ~sqlite3.SQLITE_VERSION and ~sqlite3.SQLITE_VERSION_INFO which provide information about the version of the SQLite library that was used for building the module. ~sqlite3.sqlite_version_info is now a named tuple. (Contributed by Serhiy Storchaka in 157470.)

ssl

  • Added support for Encrypted Client Hello <ssl-encrypted-client-hello> (ECH) with OpenSSL 4.0 or later. Clients can supply an ECHConfigList discovered through DNS, for example using a third-party library such as dnspython to query HTTPS or SVCB records, as the ech_config argument of ~ssl.SSLContext.wrap_socket and ~ssl.SSLContext.wrap_bio or with ~ssl.SSLContext.set_ech_config. Servers can use ~ssl.SSLContext.load_ech_keys to load ECH private keys and their configurations from PEM files. (Contributed by Peter Bierma and Iain R. Learmonth in 89730.)
  • ssl.OPENSSL_VERSION_INFO is now a named tuple. Added ~ssl.OPENSSL_API_VERSION_INFO which provides information about the version of the OpenSSL library that was used for building the module. (Contributed by Serhiy Storchaka in 157513.)
  • The ssl module now rejects TLS certificate chains that contain one of a set of compromised eMudhra intermediate CA certificates. The affected certificates are identified by their public key, so cross-signed copies that reuse the same key are blocked as well. It is a defense-in-depth backstop that applies even when an application supplies its own root CA bundle, and regardless of whether the platform trust store has removed trust. (Contributed by Christian Heimes in 160040 for 2026-115557.)

symtable

  • symtable.symtable now accepts an AST object, like the built-in compile. (Contributed by Serhiy Storchaka in 153844.)
  • Inlined list, set and dict comprehensions (709) are now represented as their own symbol table entries, of type ~symtable.SymbolTableType.INLINED_COMPREHENSION. Each such entry is a lexical child of the enclosing scope and records the comprehension's own locals, cells, and free names. It does not correspond to a separate compilation unit. For names loaded only inside an inlined list, set, or dict comprehension, symtable.Symbol.is_referenced can now return True on the enclosing function. (Contributed by Irit Katriel in 124697.)

tkinter

  • Added many tkinter.ttk.Treeview methods wrapping the enhanced ttk::treeview widget commands from Tk 9.1, such as ~tkinter.ttk.Treeview.sort, ~tkinter.ttk.Treeview.search, ~tkinter.ttk.Treeview.expand, ~tkinter.ttk.Treeview.collapse, ~tkinter.ttk.Treeview.hide, ~tkinter.ttk.Treeview.unhide, and methods for cell focus, selection and tagging. The ~tkinter.ttk.Treeview.expand and ~tkinter.ttk.Treeview.collapse methods (without recursion) also work on Tk older than 9.1. (Contributed by Serhiy Storchaka in 151910.)
  • Added constants ~tkinter.TCL_VERSION_INFO and ~tkinter.TK_VERSION_INFO, named tuples with the versions of the Tcl and Tk libraries that were used for building the module. (Contributed by Serhiy Storchaka in 157514.)
  • Added new !tkinter.Text methods ~tkinter.Text.edit_canundo and ~tkinter.Text.edit_canredo which return whether an undo or redo is possible. (Contributed by Serhiy Storchaka in 151674.)
  • Added new !tkinter.Text methods ~tkinter.Text.sync and ~tkinter.Text.pendingsync which control and report the synchronization of the displayed view with the underlying text. (Contributed by Serhiy Storchaka in 151675.)
  • Added the ttk.Style.theme_styles <tkinter.ttk.Style.theme_styles> method which returns the list of styles defined in a theme. (Contributed by Serhiy Storchaka in 151920.)
  • Added new !tkinter.Canvas methods ~tkinter.Canvas.rchars which replaces the text or coordinates of canvas items, and ~tkinter.Canvas.rotate which rotates the coordinates of canvas items. (Contributed by Serhiy Storchaka in 151876.)
  • Added a !validate method to the !tkinter.Entry and !tkinter.Spinbox widgets, which forces an evaluation of the validation command. (Contributed by Serhiy Storchaka in 151878.)
  • Added the tkinter.Menu.postcascade method, and the ~tkinter.Misc.tk_scaling and ~tkinter.Misc.tk_inactive methods which respectively query or set the display scaling factor and report the user idle time. (Contributed by Serhiy Storchaka in 151881.)
  • Added the !tk_print method to tkinter.Canvas and tkinter.Text which prints the contents of the widget using the native print dialog. It requires Tk 8.7/9.0 or newer. (Contributed by Serhiy Storchaka in 153256.)
  • Added new window-management methods ~tkinter.Misc.winfo_isdark (dark mode detection), ~tkinter.Wm.wm_iconbadge (application icon badge) and ~tkinter.Wm.wm_stackorder (toplevel stacking order). (Contributed by Serhiy Storchaka in 151874.)
  • Added the ~tkinter.Misc.tk_appname, ~tkinter.Misc.tk_useinputmethods and ~tkinter.Misc.tk_caret methods, exposing the application send name, the X Input Methods state and the input method caret location. (Contributed by Serhiy Storchaka in 151886.)
  • Added support for more options in !tkinter.PhotoImage methods: the format parameter of ~tkinter.PhotoImage.put, the metadata parameter of ~tkinter.PhotoImage.put, ~tkinter.PhotoImage.read, ~tkinter.PhotoImage.write and ~tkinter.PhotoImage.data, and the withalpha parameter of ~tkinter.PhotoImage.get. (Contributed by Serhiy Storchaka in 151890.)
  • Added the ~tkinter.PhotoImage.redither method which recalculates the dithered image when its data was supplied in pieces. (Contributed by Serhiy Storchaka in 151888.)
  • tkinter.OptionMenu now accepts arbitrary !tkinter.Menubutton options as keyword arguments, which can also override its default appearance. (Contributed by Serhiy Storchaka in 101284.)
  • The tkinter.simpledialog dialogs were modernized to match the look and feel of the native Tk dialogs. !tkinter.simpledialog.SimpleDialog and the ~tkinter.simpledialog.askinteger, ~tkinter.simpledialog.askfloat and ~tkinter.simpledialog.askstring dialogs are now built from the themed tkinter.ttk widgets instead of the classic tkinter widgets; the !tkinter.simpledialog.Dialog base class still defaults to the classic widgets for compatibility. Both !Dialog and !SimpleDialog gained a use_ttk parameter that selects between the classic Tk widgets and the themed ttk widgets. !SimpleDialog also gained bitmap and detail parameters, draws the standard icons with themed images in the ttk version, and accepts mappings of button options as buttons entries. (Contributed by Serhiy Storchaka in 59396.)
  • The !tkinter.filedialog.FileDialog dialog and its !tkinter.filedialog.LoadFileDialog and !tkinter.filedialog.SaveFileDialog subclasses are now built from the themed tkinter.ttk widgets by default instead of the classic tkinter widgets, and gained a use_ttk parameter to select between them. (Contributed by Serhiy Storchaka in 59396.)
  • Added the tkinter.systray module which provides the ~tkinter.systray.SysTrayIcon class as an interface to the system tray icon and the ~tkinter.systray.notify function which sends a desktop notification. They require Tk 8.7/9.0 or newer. (Contributed by Serhiy Storchaka in 153259.)
  • tkinter.scrolledtext.ScrolledText gained a use_ttk parameter to use the themed tkinter.ttk frame and scroll bar instead of the classic tkinter widgets. (Contributed by Serhiy Storchaka in 59396.)
  • tkinter.font.Font can now wrap a font description without creating a new named font, by passing it as font with exists=True and no name. This avoids a loss of precision in ~tkinter.font.Font.actual, ~tkinter.font.Font.measure and ~tkinter.font.Font.metrics. (Contributed by Serhiy Storchaka in 143990.)
  • Added the tkinter.fontchooser module which provides the ~tkinter.fontchooser.FontChooser class as an interface to the native font selection dialog. (Contributed by Serhiy Storchaka in 72880.)
  • Values of several Tcl object types returned by tkinter are now converted to the corresponding Python type instead of being wrapped in a !_tkinter.Tcl_Obj: index, window, nsName and parsedVarName objects to str, and pixel screen distances with no unit suffix to int or float. (Contributed by Serhiy Storchaka in 153513.)

turtle

  • Translations of the turtle docstrings are now distributed on PyPI in the turtle-translations package. Additionally, added the PYTHON_TURTLE_LANG environment variable to select the language. (Contributed by Stan Ulbrych in 156360.)

unicodedata

  • The Unicode database has been updated to Unicode 18.0.0.

uuid

  • uuid.UUID, uuid.uuid4, and uuid.uuid7 now have C implementations, with a pure Python fallback. In microbenchmarks, UUIDv4 and UUIDv7 generation are approximately 15 and 30 times faster, respectively, than the pure Python implementations. UUID construction and string conversion are also faster; see whatsnew316-uuid-optimizations for the benchmark results. (Contributed by Yury Selivanov in 139123.)

xml

  • Add support for multiple multi-byte encodings in the XML parser <xml.parsers.expat>: "cp932", "cp949", "cp950", "Big5", "EUC-JP", "GB2312", "GBK", "johab", and "Shift_JIS". Add partial support (only BMP characters) for multi-byte encodings "Big5-HKSCS", "EUC_JIS-2004", "EUC_JISX0213", "Shift_JIS-2004", "Shift_JISX0213", "utf-8-sig" and non-standard aliases like "UTF8" (without hyphen). The parser now raises ValueError for known unsupported multi-byte encodings such as "ISO-2022-JP" or "raw-unicode-escape" instead of failing later, when encountering non-ASCII data. (Contributed by Serhiy Storchaka in 62259.)
  • xml.dom.minidom now conforms closer to the DOM Level 1 specification. It checks names passed to the factory methods, rejects inserting or setting a node created by another document or making a node a descendant of itself, rejects a second element or document type child of a document and a notation child of a document fragment, reports attributes defaulted in the DTD and whether an attribute was given in the start tag, and implements !EntityReference nodes and !Document.createEntityReference. (Contributed by Jason Orendorff and Serhiy Storchaka in 44871.)
  • xml.dom.minidom now validates namespaces in the factory methods ~xml.dom.Document.createElementNS, ~xml.dom.Document.createAttributeNS and ~xml.dom.Element.setAttributeNS. ~xml.dom.NamespaceErr is now raised for a malformed qualified name, for a prefix with an empty namespace, and for illegal use of the xml and xmlns prefixes. (Contributed by Serhiy Storchaka in 156665.)
  • ~xml.dom.minidom.Node.toprettyxml in xml.dom.minidom and ~xml.etree.ElementTree.indent in xml.etree.ElementTree no longer add whitespace inside an element which is marked with xml:space="preserve" or which contains text. !toprettyxml also takes into account the content model declared in the DTD. (Contributed by Serhiy Storchaka in 81623.)
  • Add !GetSpecifiedAttributeCount method to the XML parser <xml.parsers.expat> objects. It tells how many of the reported attributes were given in the start tag rather than defaulted from the DTD. (Contributed by Jason Orendorff and Serhiy Storchaka in 44871.)
  • The ~xml.dom.minidom.Node.writexml, ~xml.dom.minidom.Node.toxml and ~xml.dom.minidom.Node.toprettyxml methods now work for !DocumentFragment nodes in xml.dom.minidom. (Contributed by Serhiy Storchaka in 54092.)
  • ~xml.etree.ElementTree.XMLPullParser and ~xml.etree.ElementTree.iterparse now support the target parameter. The reported object is the value returned by the corresponding method of the target, so a large document can be parsed incrementally without building a tree for it. (Contributed by Serhiy Storchaka in 63102.)
  • Added constant ~xml.parsers.expat.VERSION_INFO in the XML parser <xml.parsers.expat> module, which provides information about the version of the Expat library that was used for building the module. ~xml.parsers.expat.version_info is now a named tuple. (Contributed by Serhiy Storchaka in 157473.)

zipfile

  • Add ZipFile.remove() <zipfile.ZipFile.remove> to remove a member from an archive's central directory, and ZipFile.repack() <zipfile.ZipFile.repack> to reclaim the space used by the local file entries of removed members. (Contributed by Danny Lin in 51067.)

zlib

  • Added constants ~zlib.ZLIB_VERSION_INFO, ~zlib.zlib_version_info, and ~zlib.ZLIBNG_VERSION_INFO, which provide information about the version of the zlib and the zlib-ng libraries in use. Added ~zlib.zlib_version as an alias of ~zlib.ZLIB_RUNTIME_VERSION. (Contributed by Serhiy Storchaka in 117404.)

Optimizations

datetime and time

  • datetime.datetime.strptime, datetime.date.strptime, datetime.time.strptime, and time.strptime are faster for common numeric formats such as %Y-%m-%d and %Y-%m-%d %H:%M:%S. (Contributed by Paul Ganssle in 158252, based on work by Stan Ulbrych.)

re

  • Character class escapes (\d, \D, \s, \S, \w and \W) outside a character set, and character sets containing a single such escape (such as [\d] or [^\s]), are now compiled to a single CATEGORY opcode instead of being wrapped in an IN block. This speeds up matching of patterns such as \d+ and reduces the size of the compiled byte code. (Contributed by Serhiy Storchaka in 152033 and Pieter Eendebak in 152056.)

uuid

  • C implementations accelerate UUID generation, construction, and string conversion. Microbenchmarks comparing the C and pure Python implementations measured the following speedups:

    Operation Speedup
    uuid.uuid4 generation 15.01x
    uuid.uuid7 generation 29.64x
    uuid.UUID from a string 6.76x
    uuid.UUID from bytes 5.16x
    str conversion of a UUID 6.66x

    (Contributed by Yury Selivanov in 139123.)

Removed

annotationlib

  • The !annotationlib.ForwardRef._evaluate method which has been deprecated since Python 3.14. Use annotationlib.ForwardRef.evaluate or typing.evaluate_forward_ref instead.

array

  • The 'u' format code (wchar_t) which has been deprecated in documentation since Python 3.3 and at runtime since Python 3.13. Use 'w' format code instead (Py_UCS4, always 4 bytes).

asyncio

  • The !asyncio.iscoroutinefunction which has been deprecated since Python 3.14. Use inspect.iscoroutinefunction instead.
  • The event loop policy system, including the !asyncio.AbstractEventLoopPolicy, !asyncio.DefaultEventLoopPolicy, !asyncio.WindowsSelectorEventLoopPolicy and !asyncio.WindowsProactorEventLoopPolicy classes and the !asyncio.get_event_loop_policy and !asyncio.set_event_loop_policy functions, which have been deprecated since Python 3.14. Use asyncio.run or asyncio.Runner with a custom loop_factory instead.

functools

  • Remove support for calling the Python implementation of functools.reduce with function or sequence as keyword arguments, deprecated since Python 3.14. These parameters are now positional-only. (Contributed by Nikita Sobolev in 149537.)

logging

  • Remove support for custom logging handlers with the strm argument, deprecated since Python 3.14. Define handlers with the stream argument instead. (Contributed by Nikita Sobolev in 149598.)

mimetypes

  • Remove support for undotted extensions in mimetypes.MimeTypes.add_type, deprecated since Python 3.14. Valid extensions must now start with a '.' or be empty; undotted extensions raise a ValueError. (Contributed by Nikita Sobolev in 149720.)

shutil

  • The !ExecError exception which has been deprecated since Python 3.14. It has not been used by any function in !shutil since Python 3.4. (Contributed by Stan Ulbrych in 149567.)

symtable

  • The !symtable.Class.get_methods method which has been deprecated since Python 3.14.

sys

  • The !_enablelegacywindowsfsencoding function which has been deprecated since Python 3.13. Use the PYTHONLEGACYWINDOWSFSENCODING environment variable instead. (Contributed by Stan Ulbrych in 149595.)

sysconfig

  • The !sysconfig.expand_makefile_vars function which has been deprecated since Python 3.14. Use the vars argument of sysconfig.get_paths instead. (Contributed by Stan Ulbrych in 149499.)

tarfile

  • The undocumented and unused !tarfile.TarFile.tarfile attribute has been deprecated since Python 3.13.

xml.etree.ElementTree

  • The undocumented !xml.etree.ElementTree.ElementTree.write_c14n method and the support of method="c14n" in ~xml.etree.ElementTree.ElementTree.write and ~xml.etree.ElementTree.tostring. They never worked and always raised ValueError. Use xml.etree.ElementTree.canonicalize instead. (Contributed by Daniel Hillier in 118318.)

Deprecated

New deprecations

  • abc:
    • abc.abstractclassmethod, abc.abstractstaticmethod, and abc.abstractproperty, soft-deprecated since Python 3.3, now raise a DeprecationWarning. These classes will be removed in Python 3.21; use abc.abstractmethod with classmethod, staticmethod, and property respectively instead.
  • ast:
    • Classes slice, Index, ExtSlice, Suite, Param, AugLoad and AugStore, deprecated since Python 3.9, are no longer imported by from ast import * and issue a deprecation warning on use. The classes are slated for removal in Python 3.21. These types are not generated by the parser or accepted by the code generator.
    • The dims property of ast.Tuple objects, deprecated since Python 3.9, now issues a deprecation warning on use. This property is slated for removal in 3.21. Use ast.Tuple.elts instead.
  • __author__, __credits__, __copyright__ and __date__:

    • The __author__, __credits__, __copyright__ and __date__ attributes have been deprecated in these standard library modules and will be removed in Python 3.20. Attribution is preserved in the source code and the version control history.

      • bz2
      • concurrent.futures
      • inspect
      • io
      • json
      • logging
      • optparse
      • platform
      • pydoc
      • tarfile
      • tkinter.ttk
      • tokenize
      • uuid

      (Contributed by Stan Ulbrych in 159240.)

  • functools:
    • Using a return annotation to infer the dispatch type for functools.singledispatch or functools.singledispatchmethod now emits a DeprecationWarning and will raise TypeError in Python 3.21. Annotate the dispatch parameter or pass the dispatch type explicitly instead. (Contributed by Bartosz Sławecki in 143465.)
  • struct:
    • The 'F' and 'D' type codes, soft-deprecated since Python 3.15, are now deprecated. These codes will be removed in Python 3.21. Use the two-letter forms 'Zf' and 'Zd' instead. (Contributed by Sergey B Kirpichev in 121249.)
  • tempfile:
    • The private tempfile._TemporaryFileWrapper name is deprecated and is slated for removal in Python 3.21. Use the new public tempfile.TemporaryFileWrapper instead, which is the return type of tempfile.NamedTemporaryFile.
  • tkinter:
    • tkinter.filedialog.askopenfiles is deprecated and slated for removal in Python 3.19. Opening several files at once is error-prone, and the returned list cannot be used in a with statement. Iterate over the names returned by ~tkinter.filedialog.askopenfilenames and open them one by one instead. (Contributed by Serhiy Storchaka in 152638.)

Pending removal in Python 3.17

  • datetime:
    • ~datetime.datetime.strptime calls using a format string containing %e (day of month) without a year. This has been deprecated since Python 3.15. (Contributed by Stan Ulbrych in 70647.)
  • collections.abc:

    • collections.abc.ByteString is scheduled for removal in Python 3.17.

      Use isinstance(obj, collections.abc.Buffer) to test if obj implements the buffer protocol <bufferobjects> at runtime. For use in type annotations, either use ~collections.abc.Buffer or a union that explicitly specifies the types your code supports (e.g., bytes | bytearray | memoryview).

      !ByteString was originally intended to be an abstract class that would serve as a supertype of both bytes and bytearray. However, since the ABC never had any methods, knowing that an object was an instance of !ByteString never actually told you anything useful about the object. Other common buffer types such as memoryview were also never understood as subtypes of !ByteString (either at runtime or by static type checkers).

      See PEP 688 <688#current-options> for more details. (Contributed by Shantanu Jain in 91896.)

  • encodings:
    • Passing non-ascii encoding names to encodings.normalize_encoding is deprecated and scheduled for removal in Python 3.17. (Contributed by Stan Ulbrych in 136702.)
  • profile:
    • The !profile module is deprecated and will be removed in Python 3.17. Use profiling.tracing instead, which provides a compatible API. See 799 for details. (Contributed by Pablo Galindo and László Kiss Kollár in 138122.)
  • webbrowser:
    • !webbrowser.MacOSXOSAScript is deprecated in favour of !webbrowser.MacOS. (137586)
  • typing:

    • Before Python 3.14, old-style unions were implemented using the private class typing._UnionGenericAlias. This class is no longer needed for the implementation, but it has been retained for backward compatibility, with removal scheduled for Python 3.17. Users should use documented introspection helpers like typing.get_origin and typing.get_args instead of relying on private implementation details.

    • typing.ByteString, deprecated since Python 3.9, is scheduled for removal in Python 3.17.

      Use isinstance(obj, collections.abc.Buffer) to test if obj implements the buffer protocol <bufferobjects> at runtime. For use in type annotations, either use ~collections.abc.Buffer or a union that explicitly specifies the types your code supports (e.g., bytes | bytearray | memoryview).

      !ByteString was originally intended to be an abstract class that would serve as a supertype of both bytes and bytearray. However, since the ABC never had any methods, knowing that an object was an instance of !ByteString never actually told you anything useful about the object. Other common buffer types such as memoryview were also never understood as subtypes of !ByteString (either at runtime or by static type checkers).

      See PEP 688 <688#current-options> for more details. (Contributed by Shantanu Jain in 91896.)

  • tkinter:
    • The !tkinter.Variable methods !trace_variable, !trace (an alias of !trace_variable), !trace_vdelete and !trace_vinfo, deprecated since Python 3.14, are scheduled for removal in Python 3.17. Use !trace_add, !trace_remove and !trace_info instead. (Contributed by Serhiy Storchaka in 120220.)

Pending removal in Python 3.18

  • No longer accept a boolean value when a file descriptor is expected. (Contributed by Serhiy Storchaka in 82626.)

  • decimal:

    • The non-standard and undocumented ~decimal.Decimal format specifier 'N', which is only supported in the !decimal module's C implementation, has been deprecated since Python 3.13. (Contributed by Serhiy Storchaka in 89902.)

  • Deprecations defined by 829:

    • import lines in {name}.pth files are silently ignored.

    (Contributed by Barry Warsaw in 148641.)

Pending removal in Python 3.19

  • builtins:
    • Passing a complex number as the real or imag argument in the complex constructor is now deprecated; it should only be passed as a single positional argument. (Contributed by Serhiy Storchaka in 109218.)
  • ctypes:
    • Implicitly switching to the MSVC-compatible struct layout by setting ~ctypes.Structure._pack_ but not ~ctypes.Structure._layout_ on non-Windows platforms.
  • hashlib:

    • In hash function constructors such as ~hashlib.new or the direct hash-named constructors such as ~hashlib.md5 and ~hashlib.sha256, their optional initial data parameter could also be passed a keyword argument named data= or string= in various !hashlib implementations.

      Support for the string keyword argument name is now deprecated and slated for removal in Python 3.19.

      Before Python 3.13, the string keyword parameter was not correctly supported depending on the backend implementation of hash functions. Prefer passing the initial data as a positional argument for maximum backwards compatibility.

  • http.cookies:
    • http.cookies.Morsel.js_output is deprecated and will be removed in Python 3.19.
    • http.cookies.BaseCookie.js_output is deprecated and will be removed in Python 3.19.
  • imaplib:

    • Altering IMAP4.file <imaplib.IMAP4.file> is now deprecated and slated for removal in Python 3.19. This property is now unused and changing its value does not automatically close the current file.

      Before Python 3.14, this property was used to implement the corresponding read() and readline() methods for ~imaplib.IMAP4 but this is no longer the case since then.

  • tkinter:
    • tkinter.filedialog.askopenfiles has been deprecated since Python 3.16. Iterate over the names returned by ~tkinter.filedialog.askopenfilenames and open them one by one instead. (Contributed by Serhiy Storchaka in 152638.)

Pending removal in Python 3.20

  • Calling the __new__() method of struct.Struct without the format argument is deprecated and will be removed in Python 3.20. Calling ~object.__init__ method on initialized ~struct.Struct objects is deprecated and will be removed in Python 3.20.

    (Contributed by Sergey B Kirpichev and Serhiy Storchaka in 143715.)

  • The __version__, version and VERSION attributes have been deprecated in these standard library modules and will be removed in Python 3.20. Use sys.version_info instead.

    • argparse
    • csv
    • ctypes
    • !ctypes.macholib
    • decimal (use decimal.SPEC_VERSION instead)
    • http.server
    • imaplib
    • ipaddress
    • json
    • logging (__date__ also deprecated)
    • optparse
    • pickle
    • platform
    • re
    • socketserver
    • tabnanny
    • tarfile
    • tkinter.font
    • tkinter.ttk
    • wsgiref.simple_server
    • xml.etree.ElementTree
    • !xml.sax.expatreader
    • xml.sax.handler
    • zlib

    (Contributed by Hugo van Kemenade and Stan Ulbrych in 76007.)

  • The __author__, __credits__, __copyright__ and __date__ attributes have been deprecated in these standard library modules and will be removed in Python 3.20. Attribution is preserved in the source code and the version control history.

    • bz2
    • concurrent.futures
    • inspect
    • io
    • json
    • logging
    • optparse
    • platform
    • pydoc
    • tarfile
    • tkinter.ttk
    • tokenize
    • uuid

    (Contributed by Stan Ulbrych in 159240.)

  • Deprecations defined by 829:

    • Warnings are produced for import lines found in {name}.pth files.
    • {name}.pth files are no longer decoded in the locale encoding by default. They MUST be encoded in utf-8-sig.

    (Contributed by Barry Warsaw in 148641.)

  • ast:

    • Creating instances of abstract AST nodes (such as ast.AST or !ast.expr) is deprecated and will raise an error in Python 3.20.

  • typing:

    • It is deprecated to call isinstance and issubclass checks on protocol classes that were not explicitly decorated with !runtime_checkable but that inherit from a runtime-checkable protocol class. This will raise a TypeError in Python 3.20.

      (Contributed by Bartosz Sławecki in 132604.)

Pending removal in Python 3.21

  • abc

    • Soft-deprecated since Python 3.3 abc.abstractclassmethod, abc.abstractstaticmethod, and abc.abstractproperty now raise a DeprecationWarning. These classes will be removed in Python 3.21, instead use abc.abstractmethod with classmethod, staticmethod, and property respectively.

  • ast:

    • Classes slice, Index, ExtSlice, Suite, Param, AugLoad and AugStore, will be removed in Python 3.21. These types are not generated by the parser or accepted by the code generator.
    • The dims property of ast.Tuple will be removed in Python 3.21. Use the ast.Tuple.elts property instead.

  • functools:

    • Using a return annotation to infer the dispatch type for functools.singledispatch or functools.singledispatchmethod is deprecated since Python 3.16 and will raise TypeError in Python 3.21. Annotate the dispatch parameter or pass the dispatch type explicitly instead. (Contributed by Bartosz Sławecki in 143465.)

  • struct:

    • Soft-deprecated since Python 3.15, using 'F' and 'D' type codes are now deprecated. These codes will be removed in Python 3.21. Use instead two-letter forms 'Zf' and 'Zd'.

  • tempfile:

    • tempfile._TemporaryFileWrapper will be removed in Python 3.21. Use the public tempfile.TemporaryFileWrapper instead.

Pending removal in future versions

The following APIs will be removed in the future, although there is currently no date scheduled for their removal.

  • argparse:
    • Nesting argument groups and nesting mutually exclusive groups are deprecated.
    • Passing the undocumented keyword argument prefix_chars to ~argparse.ArgumentParser.add_argument_group is now deprecated.
    • The argparse.FileType type converter is deprecated.
  • builtins:
    • Generators: throw(type, exc, tb) and athrow(type, exc, tb) signature is deprecated: use throw(exc) and athrow(exc) instead, the single argument signature.
    • Currently Python accepts numeric literals immediately followed by keywords, for example 0in x, 1or x, 0if 1else 2. It allows confusing and ambiguous expressions like [0x1for x in y] (which can be interpreted as [0x1 for x in y] or [0x1f or x in y]). A syntax warning is raised if the numeric literal is immediately followed by one of keywords and, else, for, if, in, is and or. In a future release it will be changed to a syntax error. (87999)
    • Support for __index__() and __int__() method returning non-int type: these methods will be required to return an instance of a strict subclass of int.
    • Support for __float__() method returning a strict subclass of float: these methods will be required to return an instance of float.
    • Support for __complex__() method returning a strict subclass of complex: these methods will be required to return an instance of complex.
  • calendar: calendar.January and calendar.February constants are deprecated and replaced by calendar.JANUARY and calendar.FEBRUARY. (Contributed by Prince Roshan in 103636.)
  • codecs: use open instead of codecs.open. (133038)
  • !codeobject.co_lnotab: use the codeobject.co_lines method instead.
  • datetime:
    • ~datetime.datetime.utcnow: use datetime.datetime.now(tz=datetime.UTC).
    • ~datetime.datetime.utcfromtimestamp: use datetime.datetime.fromtimestamp(timestamp, tz=datetime.UTC).
  • gettext: Plural value must be an integer.
  • importlib:
    • ~importlib.util.cache_from_source debug_override parameter is deprecated: use the optimization parameter instead.
  • importlib.metadata:
    • EntryPoints tuple interface.
    • Implicit None on return values.
  • logging: the warn() method has been deprecated since Python 3.3, use ~logging.warning instead.
  • mailbox: Use of StringIO input and text mode is deprecated, use BytesIO and binary mode instead.
  • os: Calling os.register_at_fork in a multi-threaded process.
  • os.path: os.path.commonprefix is deprecated, use os.path.commonpath for path prefixes. The os.path.commonprefix function is being deprecated due to having a misleading name and module. The function is not safe to use for path prefixes despite being included in a module about path manipulation, meaning it is easy to accidentally introduce path traversal vulnerabilities into Python programs by using this function.
  • !pydoc.ErrorDuringImport: A tuple value for exc_info parameter is deprecated, use an exception instance.
  • re: More strict rules are now applied for numerical group references and group names in regular expressions. Only sequence of ASCII digits is now accepted as a numerical reference. The group name in bytes patterns and replacement strings can now only contain ASCII letters and digits and underscore. (Contributed by Serhiy Storchaka in 91760.)
  • shutil: ~shutil.rmtree's onerror parameter is deprecated in Python 3.12; use the onexc parameter instead.
  • ssl options and protocols:
    • ssl.SSLContext without protocol argument is deprecated.
    • ssl.SSLContext: ~ssl.SSLContext.set_npn_protocols and !selected_npn_protocol are deprecated: use ALPN instead.
    • ssl.OP_NO_SSL* options
    • ssl.OP_NO_TLS* options
    • ssl.PROTOCOL_SSLv3
    • ssl.PROTOCOL_TLS
    • ssl.PROTOCOL_TLSv1
    • ssl.PROTOCOL_TLSv1_1
    • ssl.PROTOCOL_TLSv1_2
    • ssl.TLSVersion.SSLv3
    • ssl.TLSVersion.TLSv1
    • ssl.TLSVersion.TLSv1_1
  • threading methods:
    • !threading.Condition.notifyAll: use ~threading.Condition.notify_all.
    • !threading.Event.isSet: use ~threading.Event.is_set.
    • !threading.Thread.isDaemon, threading.Thread.setDaemon: use threading.Thread.daemon attribute.
    • !threading.Thread.getName, threading.Thread.setName: use threading.Thread.name attribute.
    • !threading.currentThread: use threading.current_thread.
    • !threading.activeCount: use threading.active_count.
  • typing.Text (92332).
  • The internal class typing._UnionGenericAlias is no longer used to implement typing.Union. To preserve compatibility with users using this private class, a compatibility shim will be provided until at least Python 3.17. (Contributed by Jelle Zijlstra in 105499.)
  • unittest.IsolatedAsyncioTestCase: it is deprecated to return a value that is not None from a test case.
  • urllib.parse deprecated functions: ~urllib.parse.urlparse instead
    • splitattr()
    • splithost()
    • splitnport()
    • splitpasswd()
    • splitport()
    • splitquery()
    • splittag()
    • splittype()
    • splituser()
    • splitvalue()
    • to_bytes()
  • wsgiref: SimpleHandler.stdout.write() should not do partial writes.
  • xml.etree.ElementTree: Testing the truth value of an ~xml.etree.ElementTree.Element is deprecated. In a future release it will always return True. Prefer explicit len(elem) or elem is not None tests instead.
  • sys._clear_type_cache is deprecated: use sys._clear_internal_caches instead.

Porting to Python 3.16

This section lists previously described changes and other bugfixes that may require changes to your code.

  • In tkinter, the name parameter of the ~tkinter.Misc.wait_variable, ~tkinter.Misc.setvar and ~tkinter.Misc.getvar methods and the value parameter of !setvar are now required. Calling these methods without them, which formerly defaulted to 'PY_VAR' and '1', now raises TypeError. (Contributed by Serhiy Storchaka in 152587.)
  • uuid.UUID now consistently raises TypeError when the bytes or bytes_le argument is not a bytes object. Previously, some invalid inputs raised AssertionError or ValueError. (Contributed by Yury Selivanov in 139123.)
  • xml.dom.minidom now raises ~xml.dom.InvalidCharacterErr for a name which is not a valid XML name, ~xml.dom.WrongDocumentErr for inserting or setting a node created by another document, ~xml.dom.InuseAttributeErr for setting an attribute node which belongs to another element, and ~xml.dom.HierarchyRequestErr for inserting a node into itself or its descendant. It also rejects a second element or document type child of a document and a notation child of a document fragment. Such operations formerly succeeded and produced an invalid document or an endless loop. On the other hand, !Element.removeAttribute and !Element.removeAttributeNS no longer raise ~xml.dom.NotFoundErr if there is no matching attribute. Attributes defaulted in the DTD are no longer omitted when parsing. (Contributed by Jason Orendorff and Serhiy Storchaka in 44871.)
  • xml.dom.minidom now raises ~xml.dom.NamespaceErr for a malformed qualified name, for a prefix with an empty namespace, and for illegal use of the xml and xmlns prefixes. Such operations formerly succeeded and produced an invalid document. (Contributed by Serhiy Storchaka in 156665.)
  • ~xml.dom.minidom.Node.toprettyxml in xml.dom.minidom and ~xml.etree.ElementTree.indent in xml.etree.ElementTree no longer add whitespace inside an element which is marked with xml:space="preserve" or which contains text, because this changed the content of the element. !toprettyxml also takes into account the content model declared in the DTD. (Contributed by Serhiy Storchaka in 81623.)
  • On Windows, seeking a pipe now fails instead of silently appearing to succeed: os.lseek and ~io.IOBase.seek raise OSError, and ~io.IOBase.seekable returns False. As a consequence, opening a pipe in a read-write binary mode ('r+b' or 'w+b') now raises io.UnsupportedOperation unless buffering is disabled. (Contributed by An Long in 86768.)

Build changes

  • Remove the bundled copy of the libmpdec decimal library from the CPython source tree to simplify maintenance and updates. The decimal module will now unconditionally use the system's libmpdec decimal library. Also remove the now unused !--with-system-libmpdec configure flag. This change has no impact on binary releases of Python, which have been built against a separate copy of libmpdec for the past several releases.

    (Contributed by Sergey B Kirpichev in 115119.)

  • Add a --with-build-details-suffix configure flag to allow Linux distributions that co-install multiple versions of Python in the same tree to avoid build-details.json clashes.

    (Contributed by Stefano Rivera in 131372.)

  • Add the --with-curses configure option to select the curses backend for the curses and curses.panel modules. In addition to ncursesw and ncurses, it can now build against the system's native curses library (for example on NetBSD or Solaris), with wide-character support when the library provides it. --without-curses excludes the modules from the build.

    (Contributed by Serhiy Storchaka in 136687.)

  • Add the --with-zlib and --with-bzip2 configure options to select the zlib and bzip2 implementations used to build the zlib and bz2 modules. In addition to the default system libraries, they can build against zlib-rs and libbzip2-rs, and --with-zlib=zlib-ng verifies that the detected zlib is zlib-ng. --without-zlib and --without-bzip2 exclude the modules from the build.

    (Contributed by Stan Ulbrych in 139314.)

  • Add the --with-hwaddress-sanitizer configure option to build with HWAddressSanitizer.

    (Contributed by Stan Ulbrych, Florian Mayer and AnnaAr321 in 156049.)

C API changes

New features

  • Add Py_HashBuffer to the limited C API. (Contributed by Victor Stinner in 148233.)
  • Add Py_SETREF and Py_XSETREF macros to the limited C API. (Contributed by Nathan Goldbaum and Victor Stinner in 158842.)

Porting to Python 3.16

  • PyType_ClearCache is now a no-op as the type cache is now implemented per-type. It still returns the current version tag.
  • Functions reading marshalled data from a FILE*, such as PyMarshal_ReadObjectFromFile, now raise OSError for I/O errors and KeyboardInterrupt for interrupted reading, instead of EOFError. (Contributed by Serhiy Storchaka in 155907.)
  • PyMarshal_WriteLongToFile and PyMarshal_WriteObjectToFile now set the error indicator for I/O errors and interrupted writing, instead of ignoring them. PyMarshal_WriteObjectToFile now also sets the error indicator if the value cannot be marshalled. (Contributed by Serhiy Storchaka in 155907.)
  • _Py_c_abs no longer sets errno to zero on success, but rather leaves it unchanged. (Contributed by Sergey B Kirpichev in 155526.)

Deprecated C APIs

  • PyGen_New, PyGen_NewWithQualName, PyCoro_New, and PyAsyncGen_New are deprecated. They are scheduled for removal in 3.18.
  • Deprecate PyInitConfig_GetExitCode, PyStatus_Exit, and PyStatus_IsExit functions, and PyStatus.exitcode member. Py_InitializeFromInitConfig and Py_InitializeFromConfig can no longer return an exit code. Instead, if a command line option wants to exit Python, the option is now processed in Py_RunMain. (Contributed by Victor Stinner in 158080.)
  • PyModule_GetFilename is no longer deprecated, but using PyModule_GetFilenameObject instead is still recommended. (Contributed by Victor Stinner in 154757.)
  • Soft deprecate functions modifying Unicode strings: PyUnicode_New, PyUnicode_CopyCharacters, PyUnicode_Fill, PyUnicode_Resize, PyUnicode_WRITE and PyUnicode_WriteChar. Use the safer PyUnicodeWriter API instead. (Contributed by Victor Stinner in 157710.)

Pending removal in Python 3.18

  • The following private functions are deprecated and planned for removal in Python 3.18:

    • !_PyBytes_Join: use PyBytes_Join.
    • !_PyDict_GetItemStringWithError: use PyDict_GetItemStringRef.
    • !_PyDict_Pop(): use PyDict_Pop.
    • !_PyLong_Sign(): use PyLong_GetSign.
    • !_PyLong_FromDigits and !_PyLong_New: use PyLongWriter_Create.
    • !_PyThreadState_UncheckedGet: use PyThreadState_GetUnchecked.
    • !_PyUnicode_AsString: use PyUnicode_AsUTF8.
    • !_PyUnicodeWriter_Init: replace _PyUnicodeWriter_Init(&writer) with writer = PyUnicodeWriter_Create(0) <PyUnicodeWriter_Create>.
    • !_PyUnicodeWriter_Finish: replace _PyUnicodeWriter_Finish(&writer) with PyUnicodeWriter_Finish(writer) <PyUnicodeWriter_Finish>.
    • !_PyUnicodeWriter_Dealloc: replace _PyUnicodeWriter_Dealloc(&writer) with PyUnicodeWriter_Discard(writer) <PyUnicodeWriter_Discard>.
    • !_PyUnicodeWriter_WriteChar: replace _PyUnicodeWriter_WriteChar(&writer, ch) with PyUnicodeWriter_WriteChar(writer, ch) <PyUnicodeWriter_WriteChar>.
    • !_PyUnicodeWriter_WriteStr: replace _PyUnicodeWriter_WriteStr(&writer, str) with PyUnicodeWriter_WriteStr(writer, str) <PyUnicodeWriter_WriteStr>.
    • !_PyUnicodeWriter_WriteSubstring: replace _PyUnicodeWriter_WriteSubstring(&writer, str, start, end) with PyUnicodeWriter_WriteSubstring(writer, str, start, end) <PyUnicodeWriter_WriteSubstring>.
    • !_PyUnicodeWriter_WriteASCIIString: replace _PyUnicodeWriter_WriteASCIIString(&writer, str, size) with PyUnicodeWriter_WriteASCII(writer, str, size) <PyUnicodeWriter_WriteASCII>.
    • !_PyUnicodeWriter_WriteLatin1String: replace _PyUnicodeWriter_WriteLatin1String(&writer, str, size) by decoding the string with PyUnicode_DecodeLatin1, writing the result with PyUnicodeWriter_WriteStr, and releasing the result with Py_DECREF.
    • !_PyUnicodeWriter_Prepare: (no replacement).
    • !_PyUnicodeWriter_PrepareKind: (no replacement).
    • !_Py_HashPointer: use Py_HashPointer.
    • !_Py_fopen_obj: use Py_fopen.
    • PyGen_New: (no replacement).
    • PyGen_NewWithQualName: (no replacement).
    • PyCoro_New: (no replacement).
    • PyAsyncGen_New: (no replacement).

    The pythoncapi-compat project can be used to get these new public functions on Python 3.13 and older. (Contributed by Victor Stinner in 128863.)

Pending removal in Python 3.19

  • 456 embedders support for the string hashing scheme definition.

Pending removal in Python 3.20

  • !_PyObject_CallMethodId, !_PyObject_GetAttrId and !_PyUnicode_FromId are deprecated since 3.15 and will be removed in 3.20. Instead, use PyUnicode_InternFromString() and cache the result in the module state, then call PyObject_CallMethod or PyObject_GetAttr. (Contributed by Victor Stinner in 141049.)
  • The cval field in PyComplexObject (128813). Use PyComplex_AsCComplex and PyComplex_FromCComplex to convert a Python complex number to/from the C Py_complex representation.
  • Macros !Py_MATH_PIl and !Py_MATH_El.

Pending removal in future versions

The following APIs are deprecated and will be removed, although there is currently no date scheduled for their removal.

  • Py_TPFLAGS_HAVE_FINALIZE: Unneeded since Python 3.8.
  • PyErr_Fetch: Use PyErr_GetRaisedException instead.
  • PyErr_NormalizeException: Use PyErr_GetRaisedException instead.
  • PyErr_Restore: Use PyErr_SetRaisedException instead.
  • PyOS_AfterFork: Use PyOS_AfterFork_Child instead.
  • PySlice_GetIndicesEx: Use PySlice_Unpack and PySlice_AdjustIndices instead.
  • PyUnicode_READY: Unneeded since Python 3.12
  • !PyErr_Display: Use PyErr_DisplayException instead.
  • !_PyErr_ChainExceptions: Use !_PyErr_ChainExceptions1 instead.
  • !PyBytesObject.ob_shash member: call PyObject_Hash instead.
  • Thread Local Storage (TLS) API:
    • PyThread_create_key: Use PyThread_tss_alloc instead.
    • PyThread_delete_key: Use PyThread_tss_free instead.
    • PyThread_set_key_value: Use PyThread_tss_set instead.
    • PyThread_get_key_value: Use PyThread_tss_get instead.
    • PyThread_delete_key_value: Use PyThread_tss_delete instead.
    • PyThread_ReInitTLS: Unneeded since Python 3.7.

Removed C APIs

  • Remove 21 global configuration variables, deprecated in Python 3.12. To initialize Python, the PyInitConfig C API <pyinitconfig_api> can be used instead. To read these variables at runtime, PyConfig_Get can be used instead. Removed variables with their PyInitConfig C API <pyinitconfig_api> and PyConfig_Get replacement:

    • !Py_DebugFlag: Use "parser_debug" instead.
    • !Py_VerboseFlag: Use "verbose" instead.
    • !Py_QuietFlag: Use "quiet" instead.
    • !Py_InteractiveFlag: Use "interactive" instead.
    • !Py_InspectFlag: Use "inspect" instead.
    • !Py_OptimizeFlag: Use "optimization_level" instead.
    • !Py_NoSiteFlag: Use "site_import" instead (opposite value).
    • !Py_BytesWarningFlag: Use "bytes_warning" instead.
    • !Py_FrozenFlag: Use "pathconfig_warnings" instead (opposite value).
    • !Py_IgnoreEnvironmentFlag: Use "use_environment" instead (opposite value).
    • !Py_DontWriteBytecodeFlag: Use "write_bytecode" instead (opposite value).
    • !Py_NoUserSiteDirectory: Use "user_site_directory" instead (opposite value).
    • !Py_UnbufferedStdioFlag: Use "buffered_stdio" instead (opposite value).
    • !Py_HashRandomizationFlag: Use "hash_seed" instead.
    • !Py_IsolatedFlag: Use "isolated" instead.
    • !Py_LegacyWindowsFSEncodingFlag: Use "legacy_windows_fs_encoding" instead.
    • !Py_LegacyWindowsStdioFlag: Use "legacy_windows_stdio" instead.
    • !Py_FileSystemDefaultEncoding, !Py_HasFileSystemDefaultEncoding: Use "filesystem_encoding" instead.
    • !Py_FileSystemDefaultEncodeErrors: Use "filesystem_errors" instead.
    • !Py_UTF8Mode: Use "utf8_mode" instead.

    (Contributed by Victor Stinner in 158001.)

  • Remove !PyEval_InitThreads function which did nothing since Python 3.7 and was deprecated since Python 3.9. (Contributed by Victor Stinner in 154757.)

  • Remove !Py_SetProgramName and !Py_SetPythonHome functions, deprecated since Python 3.11. Instead, use PyInitConfig C API <pyinitconfig_api> to set respectively "program_name" and "home". (Contributed by Victor Stinner in 158001.)

  • Remove !PySys_SetArgv and !PySys_SetArgvEx functions, deprecated since Python 3.11. Instead, set directly sys.argv, and maybe also sys.path. (Contributed by Victor Stinner in 158001.)

  • The !PyUnstable_ExecutableKinds array, as well as the macros !PyUnstable_EXECUTABLE_KIND_SKIP, !PyUnstable_EXECUTABLE_KIND_PY_FUNCTION, !PyUnstable_EXECUTABLE_KIND_BUILTIN_FUNCTION, !PyUnstable_EXECUTABLE_KIND_METHOD_DESCRIPTOR and !PyUnstable_EXECUTABLE_KINDS, were removed.

    (Contributed by Peters Bierma and Viktorin in 152105.)

这篇文档对你有帮助吗？
想留言或继续讨论？请到 Cyberforums 赛博论坛参与。 参与讨论 / 回复