Victor Stinner [Wed, 4 May 2022 11:30:23 +0000 (13:30 +0200)]
Use static inline function Py_EnterRecursiveCall() (#91988)
Currently, calling Py_EnterRecursiveCall() and
Py_LeaveRecursiveCall() may use a function call or a static inline
function call, depending if the internal pycore_ceval.h header file
is included or not. Use a different name for the static inline
function to ensure that the static inline function is always used in
Python internals for best performance. Similar approach than
PyThreadState_GET() (function call) and _PyThreadState_GET() (static
inline function).
* Rename _Py_EnterRecursiveCall() to _Py_EnterRecursiveCallTstate()
* Rename _Py_LeaveRecursiveCall() to _Py_LeaveRecursiveCallTstate()
* pycore_ceval.h: Rename Py_EnterRecursiveCall() to
_Py_EnterRecursiveCall() and Py_LeaveRecursiveCall() and
_Py_LeaveRecursiveCall()
Victor Stinner [Wed, 4 May 2022 09:59:01 +0000 (11:59 +0200)]
gh-92036: Fix gc_fini_untrack() (#92037)
Fix a crash in subinterpreters related to the garbage collector. When
a subinterpreter is deleted, untrack all objects tracked by its GC.
To prevent a crash in deallocator functions expecting objects to be
tracked by the GC, leak a strong reference to these objects on
purpose, so they are never deleted and their deallocator functions
are not called.
Inada Naoki [Wed, 4 May 2022 01:01:15 +0000 (10:01 +0900)]
gh-78214: marshal: Stabilize FLAG_REF usage (GH-8226)
Use FLAG_REF always for interned strings.
Refcounts of interned string is very unstable.
When compiling same source, refcounts of interned string in the output may be 1 or >1.
It makes FLAG_REF usage unstable.
To help reproducible build, use FLAG_REF for interned string even if refcnt(obj)==1.
Victor Stinner [Tue, 3 May 2022 18:47:29 +0000 (20:47 +0200)]
gh-91320: Fix more old-style cast warnings in C++ (#92247)
Use _Py_CAST(), _Py_STATIC_CAST() and _PyASCIIObject_CAST() in
static inline functions to fix C++ compiler warnings:
"use of old-style cast" (clang -Wold-style-cast).
test_cppext now builds the C++ test extension with -Wold-style-cast.
Toshio Kuratomi [Tue, 3 May 2022 16:38:18 +0000 (09:38 -0700)]
gh-88753: Make BooleanOptionalAction's addition of default to help more similar to other actions (#27808)
Help for other actions omit the default value if default is SUPPRESS or
already contains the special format string '%(default)'. Add those
special cases to BooleanOptionalAction's help formatting too.
Fixes https://bugs.python.org/issue44587 so that default=SUPPRESS is not
emitted.
Fixes https://bugs.python.org/issue38956 as this code will detect
whether '%(default)s' has already been specified in the help string.
Łukasz Langa [Tue, 3 May 2022 16:27:38 +0000 (18:27 +0200)]
gh-89452: Prefer gdbm-compat over ndbm (#92208)
This makes macOS gdbm provided by Homebrew not segfault through correct
selection of the linked library (-lgdbm_compat) *AND* the correct ndbm-style
header (gdbm-ndbm.h instead of the invalid ndbm.h).
Carey Metcalfe [Tue, 3 May 2022 12:18:18 +0000 (08:18 -0400)]
gh-70363: Implement `io.IOBase` interface for `SpooledTemporaryFile` (GH-29560)
Since the underlying file-like objects (either `io.BytesIO`,
or a true file object) all implement the `io.IOBase`
interface, the `SpooledTemporaryFile` should as well.
Additionally, since the underlying file object will either be an
instance of an `io.BufferedIOBase` (for binary mode) or an
`io.TextIOBase` (for text mode), methods for these classes were also
implemented.
In every case, the required methods and properties are simply delegated
to the underlying file object.
Co-authored-by: Gary Fernie <Gary.Fernie@skyscanner.net> Co-authored-by: Inada Naoki <songofacandy@gmail.com>
bpo-46415: Use f-string for ValueError in ipaddress.ip_{address,network,interface} helper functions (#30642)
`IPv*Network` and `IPv*Interface` constructors accept a 2-tuple of
(address description, netmask) as the address parameter.
When the tuple-based address is used errors are not propagated
correctly through the `ipaddress.ip_*` helper because of the %-formatting now expecting several arguments:
In [7]: ipaddress.ip_network(("192.168.100.0", "fooo"))
...
TypeError: not all arguments converted during string formatting
Compared to:
In [8]: ipaddress.IPv4Network(("192.168.100.0", "foo"))
...
NetmaskValueError: 'foo' is not a valid netmask
Use an f-string to make sure the error is always properly formatted.
Shantanu [Mon, 2 May 2022 23:08:28 +0000 (17:08 -0600)]
gh-91621: Fix typing.get_type_hints for collections.abc.Callable (#91656)
This mirrors logic in typing.get_args. The trickiness comes from how we
flatten args in collections.abc.Callable, see
https://bugs.python.org/issue42195
Victor Stinner [Mon, 2 May 2022 15:07:00 +0000 (17:07 +0200)]
gh-92135: Fix _Py_reinterpret_cast() for const (#92138)
Fix C++ compiler warnings on cast macros, like _PyObject_CAST(), when
casting a constant expression to a non constant type: use
const_cast<> in C++.
* In C++, Py_SAFE_DOWNCAST() now uses static_cast<> rather than
reinterpret_cast<>.
* Add tests to the _testcppext C++ extension.
* test_cppext no longer captures stdout in verbose mode.
bpo-36819: Fix crashes in built-in encoders with weird error handlers (GH-28593)
If the error handler returns position less or equal than the starting
position of non-encodable characters, most of built-in encoders didn't
properly re-size the output buffer. This led to out-of-bounds writes,
and segfaults.
Antony Lee [Sun, 1 May 2022 06:04:50 +0000 (08:04 +0200)]
bpo-39716: Raise on conflicting subparser names. (GH-18605)
Raise an ArgumentError when the same subparser name is added twice to an
ArgumentParser. This is consistent with the (default) behavior when the
same option string is added twice to an ArgumentParser.
(Support for `conflict_handler="resolve"` could be considered as a
followup feature, although real use cases seem even rarer than
"resolve"ing option-strings.)