python / cpython Public

…H-32110)

Co-authored-by: Jelle Zijlstra <jelle.zijlstra@gmail.com>

…izer buffers are uninitialized (GH-32129)

Automerge-Triggered-By: GH:pablogsal

Co-authored-by: Yury Selivanov <yury@edgedb.com>
Co-authored-by: Andrew Svetlov <andrew.svetlov@gmail.com>

* Fix some typos and phrasing

* Convert the `hash()` text to a link

* Remove definite article

…-31318)

* Don't deopt if refcounts are too big

* Detect more at specialization time

… it. (GH-32055)

* `PyFrame_FastToLocalsWithError` and `PyFrame_LocalsToFast` are no longer called during profile and tracing.
 (Contributed by Fabio Zadrozny)

* Make accesses to a frame's `f_locals` safe from C code, not relying on calls to `PyFrame_FastToLocals` or `PyFrame_LocalsToFast`.

* Document new `PyFrame_GetLocals` C-API function.

Shared extension on Emscripten now have suffix
``.cpython-311-wasm32-emscripten.so`` (JS loader) and
``.cpython-311-wasm32-emscripten.wasm`` (WebAssembly code).

Co-authored-by: Zachary Ware <zach@python.org>

)

The goal here is to reduce potential confusion between
`assert_type(val, type)` and `assert isinstance(val, typ)`.

The former is meant to ask a type checker to confirm a fact, the latter
is meant to tell a type checker a fact. The behaviour of the latter more
closely resembles what I'd expect from the prior phrasing of
"assert [something] to the type checker".

…Case (GH-32086)

…able (GH-32076)

…itialization (GH-31082)

…Objects (gh-32061)

We have to run "make regen-deepfreeze" before running Tools/scripts/generate-global-objects.py; otherwise we will miss any changes to global objects in deep-frozen modules (which aren't committed in the repo).  However, building $(PYTHON_FOR_FREEZE) fails if one of its source files had a global object (e.g. via _Py_ID(...)) added or removed, without generate-global-objects.py running first.  So "make regen-global-objects" would sometimes fail.

We solve this by running generate-global-objects.py before *and* after "make regen-deepfreeze". To speed things up and cut down on noise, we also avoid updating the global objects files if there are no changes to them.

https://bugs.python.org/issue46712

In a gh-32003 comment, I realized it wasn't very clear how _Py_DECLARE_STR() should be used.  This changes adds a comment to clarify.

https://bugs.python.org/issue46541

…cancel() (GH-31840)

After a long deliberation we ended up feeling that the message argument for Future.cancel(), added in 3.9, was a bad idea, so we're deprecating it in 3.11 and plan to remove it in 3.13.

Reorganize the documentation of the PyFrameObject C API.

People are testing those blocs with the default
inline_comment_prefixes of None, leading to a:

configparser.InterpolationSyntaxError: '$' must be followed by '$' or '{', found: '$ sign ($ is the only character that needs to be escaped)'

… susp-ignored.csv. (GH-32070)

…-32023)

Also increases the stack allocation when run with `python_d.exe` to account for the extra stack checks that are added.

Co-authored-by: Piet Delport
Co-authored-by: Hugo Lopes Tavares
Co-authored-by: Jelle Zijlstra <jelle.zijlstra@gmail.com>