gh-115528: Update language reference for PEP 646 (GH-121181)
To recap: the objective is to make starred expressions valid in `subscription`,
which is used for generics: `Generic[...]`, `list[...]`, etc.
What _is_ gramatically valid in such contexts? Seemingly any of the following.
(At least, none of the following throw `SyntaxError` in a 3.12.3 REPL.)
Generic[x]
Generic[*x]
Generic[*x, y]
Generic[y, *x]
Generic[x := 1]
Generic[x := 1, y := 2]
So introducting
flexible_expression: expression | assignment_expression | starred_item
end then switching `subscription` to use `flexible_expression` sorts that.
But then we need to field `yield` - for which any of the following are
apparently valid:
yield x
yield x,
yield x, y
yield *x,
yield *x, *y
Introducing a separate `yield_list` is the simplest way I've been figure out to
do this - separating out the special case of `starred_item ,`.
(cherry picked from commit 7d3497f617)
Co-authored-by: Matthew Rahtz <matthew.rahtz@gmail.com>
Co-authored-by: Jelle Zijlstra <jelle.zijlstra@gmail.com>
Using a standard library class makes this test difficult to maintain
as other tests and other parts of the stdlib may create subclasses,
which may still be alive when this test runs depending on GC timing.
(cherry picked from commit 08a467b537)
Closes GH-123242. The real criterion is that the attribute does not
exist on heap types, but I don't think we should discuss heap vs.
static types in the language reference.
(cherry picked from commit 99b23c64de)
gh-122701: Fix wording of raw strings/bytes in `lexical_analysis.rst` (GH-122702)
(cherry picked from commit ea70439bd2)
Co-authored-by: sobolevn <mail@sobolevn.me>
Co-authored-by: Adam Turner <9087854+AA-Turner@users.noreply.github.com>
Remove outdated note about instance methods from datamodel.rst (GH-122471)
(cherry picked from commit c68cb8e0c9)
Co-authored-by: Thomas Grainger <tagrain@gmail.com>
gh-114104: clarify asynchronous comprehension docs to match runtime behavior (GH-121175)
(cherry picked from commit 91313afdb3)
Co-authored-by: Danny Yang <yangdanny97@users.noreply.github.com>
doc: Mention the missing reflected special methods for all binary operations (GH-119931)
(cherry picked from commit bf5e1065f4)
Co-authored-by: Paulo Freitas <me@paulofreitas.me>
gh-120937: Reference weakref from the `__del__` documentation (GH-120940)
(cherry picked from commit 1c13b29d54)
Co-authored-by: chaen <christophe.haen@cern.ch>
Co-authored-by: Hugo van Kemenade <1324225+hugovk@users.noreply.github.com>
This behavior is rather surprising and it was not clearly specified.
(cherry picked from commit 9e0b11eb21)
Co-authored-by: Jelle Zijlstra <jelle.zijlstra@gmail.com>
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
gh-100117: Fix inaccuracy in documentation of the CodeObject's co_positions field. (GH-119364)
(cherry picked from commit 015b1fdd0a)
Co-authored-by: Irit Katriel <1055913+iritkatriel@users.noreply.github.com>
Define 'nonlocal scopes' in a way that excludes class scopes.
Rearrange the rest of the doc. Add "Programmer's note".
(cherry picked from commit 025ef7a5f7)
Co-authored-by: Terry Jan Reedy <tjreedy@udel.edu>
In Lexical Analysis f-strings section, NULL in the description
of 'literal character' means '\0'. In the format_spec grammar
production, it is wrong with that meaning and redundant if
instead interpreted as <nothing>. Remove it there.
(cherry picked from commit 4e45c6c54a)
Co-authored-by: Terry Jan Reedy <tjreedy@udel.edu>
Thanks to Pedro Arthur Duarte (pedroarthur.jedi at gmail.com) for help with this bug.
(cherry picked from commit f7455864f2)
Co-authored-by: Adorilson Bezerra <adorilson@gmail.com>
* [3.12] gh-115572: Move `codeobject.replace()` docs to the data model (GH-115631)
(cherry picked from commit 0c80da4c14)
Co-authored-by: Daler <48939169+daler-sz@users.noreply.github.com>
* Remove note about copy.replace
---------
Co-authored-by: Daler <48939169+daler-sz@users.noreply.github.com>
gh-115405: add versionadded tag for co_qualname in code objects documentation (GH-115411)
(cherry picked from commit de07941729)
Co-authored-by: Christophe Papazian <114495376+christophe-papazian@users.noreply.github.com>
A 'single tuple' means 'one tuple, of whatever length.
Remove the unneeded and slightly distracting parenthetical 'singleton' comment.
(cherry picked from commit a1332a99cf)
Co-authored-by: Terry Jan Reedy <tjreedy@udel.edu>
datamodel: Fix a typo in ``object.__init_subclass__`` (GH-111599)
(cherry picked from commit a47353d587)
Co-authored-by: InSync <122007197+InSyncWithFoo@users.noreply.github.com>
gh-101100: Fix Sphinx nitpicks in `library/inspect.rst` and `reference/simple_stmts.rst` (GH-113107)
(cherry picked from commit 4b3cb082da)
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
