* Adding MSVC C4127 suppression around Eigen includes.
* For MSVC 2015 only: also adding the C4127 suppression to test_eigen.cpp
* Copying original change from PR #3343, with extra line breaks to not run past 99 columns (our desired but currently not enforced limit).
* Add a test showing a flaw in make_key_iterator/make_value_iterator
If the iterator dereference operator returns a value rather than a
reference (and that pair also does not *contain* references),
make_key_iterator and make_value_iterator will return a reference to a
temporary, causing a segfault.
* Fix make_key_iterator/make_value_iterator for prvalue iterators
If an iterator returns a pair<T1, T2> rather than a reference to a pair
or a pair of references, make_key_iterator and make_value_iterator would
return a reference to a temporary, typically leading to a segfault. This
is because the value category of member access to a prvalue is an
xvalue, not a prvalue, so decltype produces an rvalue reference type.
Fix the type calculation to handle this case.
I also removed some decltype parentheses that weren't needed, either
because the expression isn't one of the special cases for decltype or
because decltype was only used for SFINAE. Hopefully that makes the code
a bit more readable.
Closes#3347
* Attempt a workaround for nvcc
* Add `.keys` and `.values` to bind_map
Both of these implement views (rather than just iterators), and `.items`
is also upgraded to a view. In practical terms, this allows a view to be
iterated multiple times and have its size taken, neither of which works
with an iterator.
The views implement `__len__`, `__iter__`, and the keys view implements
`__contains__`. Testing membership also works in item and value views
because Python falls back to iteration. This won't be optimal
for item values since it's linear rather than O(log n) or O(1), but I
didn't fancy trying to get all the corner cases to match Python
behaviour (tuple of wrong types, wrong length tuple, not a tuple etc).
Missing relative to Python dictionary views is `__reversed__` (only
added to Python in 3.8). Implementing that could break code that binds
custom map classes which don't provide `rbegin`/`rend` (at least without
doing clever things with SFINAE), so I've not tried.
The size increase on my system is 131072 bytes, which is rather large
(5%) but also suspiciously round (2^17) and makes me suspect some
quantisation effect.
* bind_map: support any object in __contains__
Add extra overload of `__contains__` (for both the map itself and
KeysView) which takes an arbitrary object and returns false.
* Take py::object by const reference in __contains__
To keep clang-tidy happy.
* Removing stray `py::` (detected via interactive testing in Google environment).
Co-authored-by: Ralf W. Grosse-Kunstleve <rwgk@google.com>
* Fix `pybind11::object::operator=` to be safe if `*this` is accessible from Python
* Add `custom_type_setup` attribute
This allows for custom modifications to the PyHeapTypeObject prior to
calling `PyType_Ready`. This may be used, for example, to define
`tp_traverse` and `tp_clear` functions.
* Add make_value_iterator (#3271)
* Add make_value_iterator
This is the counterpart to make_key_iterator, and will allow
implementing a `value` method in `bind_map` (although doing so is left
for a subsequent PR).
I made a few design changes to reduce copy-and-paste boilerplate.
Previously detail::iterator_state had a boolean template parameter to
indicate whether it was being used for make_iterator or
make_key_iterator. I replaced the boolean with a class that determines
how to dereference the iterator. This allows for a generic
implementation of `__next__`.
I also added the ValueType and Extra... parameters to the iterator_state
template args, because I think it was a bug that they were missing: if
make_iterator is called twice with different values of these, only the
first set has effect (because the state class is only registered once).
There is still a potential issue in that the *values* of the extra
arguments are latched on the first call, but since most policies are
empty classes this should be even less common.
* Add some remove_cv_t to appease clang-tidy
* Make iterator_access and friends take reference
For some reason I'd accidentally made it take a const value, which
caused some issues with third-party packages.
* Another attempt to remove remove_cv_t from iterators
Some of the return types were const (non-reference) types because of the
pecularities of decltype: `decltype((*it).first)` is the *declared* type
of the member of the pair, rather than the type of the expression. So if
the reference type of the iterator is `pair<const int, int> &`, then the
decltype is `const int`. Wrapping an extra set of parentheses to form
`decltype(((*it).first))` would instead give `const int &`.
This means that the existing make_key_iterator actually returns by value
from `__next__`, rather than by reference. Since for mapping types, keys
are always const, this probably hasn't been noticed, but it will affect
make_value_iterator if the Python code tries to mutate the returned
objects. I've changed things to use double parentheses so that
make_iterator, make_key_iterator and make_value_iterator should now all
return the reference type of the iterator. I'll still need to add a test
for that; for now I'm just checking whether I can keep Clang-Tidy happy.
* Add back some NOLINTNEXTLINE to appease Clang-Tidy
This is favoured over using remove_cv_t because in some cases a const
value return type is deliberate (particularly for Eigen).
* Add a unit test for iterator referencing
Ensure that make_iterator, make_key_iterator and make_value_iterator
return references to the container elements, rather than copies. The
test for make_key_iterator fails to compile on master, which gives me
confidence that this branch has fixed it.
* Make the iterator_access etc operator() const
I'm actually a little surprised it compiled at all given that the
operator() is called on a temporary, but I don't claim to fully
understand all the different value types in C++11.
* Attempt to work around compiler bugs
https://godbolt.org/ shows an example where ICC gets the wrong result
for a decltype used as the default for a template argument, and CI also
showed problems with PGI. This is a shot in the dark to see if it fixes
things.
* Make a test constructor explicit (Clang-Tidy)
* Fix unit test on GCC 4.8.5
It seems to require the arguments to the std::pair constructor to be
implicitly convertible to the types in the pair, rather than just
requiring is_constructible.
* Remove DOXYGEN_SHOULD_SKIP_THIS guards
Now that a complex decltype expression has been replaced by a simpler
nested type, I'm hoping Doxygen will be able to build it without issues.
* Add comment to explain iterator_state template params
* fix: regression in #3271
Co-authored-by: Bruce Merry <1963944+bmerry@users.noreply.github.com>
* Adding nullptr slices
Using example from #1095
Some fixes from @wjakob's review
Stop clang-tidy from complaining
New proposal for py::slice constructor
Eric's suggested changes: simplify testing; shift def's
* chore: drop MSVC pragma (hopefully unneeded)
* Apply suggestions from code review
* Add make_value_iterator
This is the counterpart to make_key_iterator, and will allow
implementing a `value` method in `bind_map` (although doing so is left
for a subsequent PR).
I made a few design changes to reduce copy-and-paste boilerplate.
Previously detail::iterator_state had a boolean template parameter to
indicate whether it was being used for make_iterator or
make_key_iterator. I replaced the boolean with a class that determines
how to dereference the iterator. This allows for a generic
implementation of `__next__`.
I also added the ValueType and Extra... parameters to the iterator_state
template args, because I think it was a bug that they were missing: if
make_iterator is called twice with different values of these, only the
first set has effect (because the state class is only registered once).
There is still a potential issue in that the *values* of the extra
arguments are latched on the first call, but since most policies are
empty classes this should be even less common.
* Add some remove_cv_t to appease clang-tidy
* Make iterator_access and friends take reference
For some reason I'd accidentally made it take a const value, which
caused some issues with third-party packages.
* Another attempt to remove remove_cv_t from iterators
Some of the return types were const (non-reference) types because of the
pecularities of decltype: `decltype((*it).first)` is the *declared* type
of the member of the pair, rather than the type of the expression. So if
the reference type of the iterator is `pair<const int, int> &`, then the
decltype is `const int`. Wrapping an extra set of parentheses to form
`decltype(((*it).first))` would instead give `const int &`.
This means that the existing make_key_iterator actually returns by value
from `__next__`, rather than by reference. Since for mapping types, keys
are always const, this probably hasn't been noticed, but it will affect
make_value_iterator if the Python code tries to mutate the returned
objects. I've changed things to use double parentheses so that
make_iterator, make_key_iterator and make_value_iterator should now all
return the reference type of the iterator. I'll still need to add a test
for that; for now I'm just checking whether I can keep Clang-Tidy happy.
* Add back some NOLINTNEXTLINE to appease Clang-Tidy
This is favoured over using remove_cv_t because in some cases a const
value return type is deliberate (particularly for Eigen).
* Add a unit test for iterator referencing
Ensure that make_iterator, make_key_iterator and make_value_iterator
return references to the container elements, rather than copies. The
test for make_key_iterator fails to compile on master, which gives me
confidence that this branch has fixed it.
* Make the iterator_access etc operator() const
I'm actually a little surprised it compiled at all given that the
operator() is called on a temporary, but I don't claim to fully
understand all the different value types in C++11.
* Attempt to work around compiler bugs
https://godbolt.org/ shows an example where ICC gets the wrong result
for a decltype used as the default for a template argument, and CI also
showed problems with PGI. This is a shot in the dark to see if it fixes
things.
* Make a test constructor explicit (Clang-Tidy)
* Fix unit test on GCC 4.8.5
It seems to require the arguments to the std::pair constructor to be
implicitly convertible to the types in the pair, rather than just
requiring is_constructible.
* Remove DOXYGEN_SHOULD_SKIP_THIS guards
Now that a complex decltype expression has been replaced by a simpler
nested type, I'm hoping Doxygen will be able to build it without issues.
* Add comment to explain iterator_state template params
* Enable defining custom __new__
* See if xfail needed
* Qualify auto self
* Unconditionally defining PYBIND11_DISABLE_NEW_STYLE_INIT_WARNING. Returning pointer from "__init__" instead of reference.
* Use new style __init__
* Simplify __new__ creation
* Reviewer suggestions
* Match indentation
Co-authored-by: Ralf W. Grosse-Kunstleve <rwgk@google.com>
* Eliminate duplicate TLS keys for loader_life_support stack
This revises the existing fix for
https://github.com/pybind/pybind11/issues/2765 in
https://github.com/pybind/pybind11/pull/3237 to reduce the amount of
TLS storage used.
The shared TLS key is stored in two different ways, depending on
`PYBIND11_INTERNALS_VERSION`. If `PYBIND11_INTERNALS_VERSION ==
4` (as is currently set), the TLS key is stored in the
`internal::shared_data` map to avoid breaking ABI compatibility. If
`PYBIND11_INTERNALS_VERSION > 4`, the TLS key is stored directly in
the `internals` struct.
* Fix test_pytypes.py::test_issue2361 failure on PyPy3.7
* Add github actions tests for unstable ABI
* Restoring `const` removed from pytypes.h in PR #3254, adding tests reflective of user code that breaks when those `const` are removed.
* clang-tidy NOLINTs (and one collateral fix).
* Inserting PYBIND11_CONST_FOR_STRICT_PLATFORMS
* Trying `defined(__APPLE__)`
* Trying again: `auto it` for strict platforms.
* Adding NOLINTNEXTLINE(bugprone-macro-parentheses), expanding comments.
* Labeling all changes with `PR #3263`, for easy reference, and to make it easy to undo these changes if we decide to do so in the future.
* Fix thread safety for pybind11 loader_life_support
Fixes issue: https://github.com/pybind/pybind11/issues/2765
This converts the vector of PyObjects to either a single void* or
a per-thread void* depending on the WITH_THREAD define.
The new field is used by each thread to construct a stack
of loader_life_support frames that can extend the life of python
objects.
The pointer is updated when the loader_life_support object is allocated
(which happens before a call) as well as on release.
Each loader_life_support maintains a set of PyObject references
that need to be lifetime extended; this is done by storing them
in a c++ std::unordered_set and clearing the references when the
method completes.
* Also update the internals version as the internal struct is no longer compatible
* Add test demonstrating threading works correctly.
It may be appropriate to run this under msan/tsan/etc.
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* Update test to use lifetime-extended references rather than
std::string_view, as that's a C++ 17 feature.
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* Make loader_life_support members private
* Update version to dev2
* Update test to use python threading rather than concurrent.futures
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* Remove unnecessary env in test
* Remove unnecessary pytest in test
* Use native C++ thread_local in place of python per-thread data structures to retain compatability
* clang-format test_thread.cpp
* Add a note about debugging the py::cast() error
* thread_test.py now propagates exceptions on join() calls.
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* remove unused sys / merge
* Update include order in test_thread.cpp
* Remove spurious whitespace
* Update comment / whitespace.
* Address review comments
* lint cleanup
* Fix test IntStruct constructor.
* Add explicit to constructor
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Aaron Gokaslan <skylion.aaron@gmail.com>
* Set __file__ constant when using eval_file
* Use const ref
* Use a move instead
* Revert
* Improve test
* Guard test with Python version
* Fix tests
* Dont support Python2 API
* Drop Python2 eval __file__ support
* Hack
* Semisupport Python2
* Take2
* Remove Python2 support
* Minor tweaks.
* Restoring tests/pybind11_tests.h version from master, removing just the comment and empty line that was added in PR #3087; those were made obsolete by the pragma cleanup that concluded with PR #3186.
* More-to-the-point test for Python 3.
* Minor tweaks.
* Restoring tests/pybind11_tests.h version from master, removing just the comment and empty line that was added in PR #3087; those were made obsolete by the pragma cleanup that concluded with PR #3186.
* [ci skip] Restoring tests/test_enum.py from master.
* fix: memory leak in cpp_function (#3228)
* add a test case to check objects are deconstructed in cpp_function
* update the test case about cpp_function
* fix the test case about cpp_function: remove "noexcept"
* Actually calling func. CHECK(stat.alive() == 2); Manually verified that the new tests fails without the change in pybind11.h
* Moving new test to test_callbacks.cpp,py, with small enhancements.
* Removing new test from test_interpreter.cpp (after it was moved to test_callbacks.cpp,py). This restores test_interpreter.cpp to the current state on master.
* Using py::detail::silence_unused_warnings(py_func); to make the intent clear.
Co-authored-by: Ralf W. Grosse-Kunstleve <rwgk@google.com>
* Trivial change to avoid (ssize_t) cast.
* Demo for safe_ssize_t idea.
* Removing safe_ssize_t.cpp (proof-of-concept code) to not upset the GHA Format workflow.
* Completing changes in pytypes.h
* New ssize_t_cast (better replacement for safe_ssize_t).
* clang-format-diff (no manual changes).
* bytes_ssize_t -Wnarrowing reproducer (see PR #2692).
* Backing out tuple(), list() ssize_t support, for compatibility with older compilers (to resolve link failures).
* Bug fix: missing `py::` for `py::ssize_t`
* Restoring tuple(), list() ssize_t support, but passing `size` by value, for compatibility with older compilers (to resolve link failures).
* Full test coverage of all functions with modified signatures.
* Use equivalent_integer for enum's Scalar decision
* Add test for char underlying enum
* Support translating bool type in enum's Scalar
* Add test for bool underlying enum
* Fix comment in test
* Switch from `PYBIND11_CPP20` macro to `PYBIND11_HAS_U8STRING`
* Refine tests
Co-authored-by: Aaron Gokaslan <skylion.aaron@gmail.com>
* Call PySys_SetArgv when initializing interpreter.
* Document argc/argv parameters in initialize_interpreter.
* Remove manual memory management from set_interpreter_argv in favor of smart pointers.
* Use size_t for indexers in set_interpreter_argv.
* Minimize macros for flow control in set_interpreter_argv.
* Fix 'unused variable' warning on Py2
* whitespace
* Define wide_char_arg_deleter outside set_interpreter_argv.
* Do sys.path workaround in C++ rather than eval.
* Factor out wchar conversion to a separate function.
* Restore widened_argv variable declaration.
* Fix undeclared widened_arg variable on some paths.
* Use delete[] to match new wchar_t[].
* Fix compiler errors
* Use PY_VERSION_HEX for a cleaner CVE-2008-5983 mode check.
* Fix typo
* Use explicit type for deleter so delete[] works cross-compiler.
* Always use PySys_SetArgvEx because pybind11 doesn't support pythons that don't include it.
* Remove pointless ternary operator.
* Use unique_ptr.reset instead of a second initialization.
* Rename add_program_dir_to_path parameter to clarify intent.
* Add defined() check before evaluating HAVE_BROKEN_MBSTOWCS.
* Apply clang-tidy fixes
* Pre-commit
* refactor: use const for set_interpreter_argv
* Try to fix const issue and allocate vector properly
* fix: copy strings on Python 2
* Applying clang-format-diff relative to master.
The only manual change is an added empty line between pybind11 and system `#include`s.
```
git diff -U0 --no-color master | python3 $HOME/clone/llvm-project/clang/tools/clang-format/clang-format-diff.py -p1 -style=file -i
```
Co-authored-by: Boris Staletic <boris.staletic@gmail.com>
Co-authored-by: Aaron Gokaslan <skylion.aaron@gmail.com>
Co-authored-by: Henry Schreiner <henryschreineriii@gmail.com>
Co-authored-by: Ralf W. Grosse-Kunstleve <rwgk@google.com>
* Add py::raise_from to enable chaining exceptions on Python 3.3+
* Use 'raise from' in initialization
* Documenting the exact base version of _PyErr_FormatVFromCause, adding back `assert`s.
Co-authored-by: Dustin Spicuzza <dustin@virtualroadside.com>
* extend msys2 CI
- add 32-bit job
- add c++11/17 c++/interface tests copied from standard ci
- add numpy/scipy
* account for padding of PartialStruct in numpy dtypes test with mingw32
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* msys2 ci: add c++14 tests
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* Removing all MSVC C4127 warning suppression pragmas.
* Removing MSVC /WX (WERROR). To get a full list of all warnings.
* Inserting PYBIND11_SILENCE_MSVC_C4127. Changing one runtime if to #if.
* Changing PYBIND11_SILENCE_MSVC_C4127 macro to use absolute namespace (for use outside pybind11 include directory).
* Restoring MSVC /WX (WERROR).
* Removing globally-scoped suppression for clang -Wunsequenced. Based on an experiment under PR #3202 it is obsolete and can simply be removed.
The only manual intervention was to remove the merge conflict markers in pybind11.h.
TODO in separate commit: complete removal of the obsolete detail/pragma_warning_block.h.
* Apply isort
* Tweak isort config
* Add env.py as a known_first_party
* Add one missing known first party
* Make config compat with older isort versions
* Add another comment
* Revert pyproject setting
* test_eval: Show example of working closure
* Extend test_eval_closure with weirder examples of closures for py::eval
Co-authored-by: Yannick Jadoul <yannick.jadoul@belgacom.net>
Co-authored-by: Aaron Gokaslan <skylion.aaron@gmail.com>
* mingw64 platform string is like mingw_xxx not "mingw"
See https://github.com/msys2/MINGW-packages/blob/master/mingw-w64-python/0099-Change-the-get_platform-method-in-sysconfig-and-dist.patch
* Mingw: Do not dllexport exceptions
This is a fix for errors like:
D:/a/pybind11/pybind11/include/pybind11/detail/common.h:735:23: error: 'dllexport' implies default visibility, but 'class pybind11::builtin_exception' has already been declared with a different visibility
735 | class PYBIND11_EXPORT builtin_exception : public std::runtime_error {
| ^~~~~~~~~~~~~~~~~
* GHA: Test Mingw64 build
* fix: avoid thin binaries on mingw
* fix: drop lto on MinGW
* Mingw64: disable PYBIND11_DEPRECATED
It trigger many warnings for unknown reasons
Co-authored-by: Henry Schreiner <henryschreineriii@gmail.com>
* Fixing `pragma warning pop` for `__INTEL_COMPILER`.
* Adding push/pop to 3 tests. Removing #878 from top of pybind11.h (it was/is only needed for 1 test).
* Trying again after CI failure, moving the push to the top of 2 tests.
* Trying more after CI failure, adding push/pop to pybind11_tests.h, constructor_stats.h.
* Moving ICC #2196 suppression to CMakeLists.txt
* Fixing condition for `pragma GCC diagnostic push` in pybind11.h
* Moving `pragma warning disable 2196` to common.h
* Revising #ifdef to be more conservative.
* Undoing insertion of notes that will hopefully soon be completely obsolete anyway.
* maint(clang-tidy) Improve code readability
* Fix minor typos
* Revert optimization that removed test case
* Fix comment formatting
* Revert another optimization to repro an issue
* Remove make_unique since it C++14 and newer only
* eformat comments
* Fix unsignedness of comparison
* Update comment
* Allow python builtins to be used as callbacks
* Try to fix pypy segfault
* Add expected fail for PyPy
* Fix typo
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* Add more info to xfail
* Add env
* Try returning false
* Try removing the move for pypy
* Fix bugs
* Try removing move
* Just keep ignoring for PyPy
* Add back xfail
* Fix ctors
* Revert change of std::move
* Change to skip
* Fix bug and edit comments
* Remove clang-tidy bugprone fix skip bug
Co-authored-by: Aaron Gokaslan <skylion.aaron@gmail.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* Create a module_internals struct
Since we now have two things that are going to be module local, it felt
correct to add a struct to manage them.
* Add local exception translators
These are added via the register_local_exception_translator function
and are then applied before the global translators
* Add unit tests to show the local exception translator works
* Fix a bug in the unit test with the string value of KeyError
* Fix a formatting issue
* Rename registered_local_types_cpp()
Rename it to get_registered_local_types_cpp() to disambiguate from the
new member of module_internals
* Add additional comments to new local exception code path
* Add a register_local_exception function
* Add additional unit tests for register_local_exception
* Use get_local_internals like get_internals
* Update documentation for new local exception feature
* Add back a missing space
* Clean-up some issues in the docs
* Remove the code duplication when translating exceptions
Separated out the exception processing into a standalone function in the
details namespace.
Clean-up some comments as per PR notes as well
* Remove the code duplication in register_exception
* Cleanup some formatting things caught by clang-format
* Remove the templates from exception translators
But I added a using declaration to alias the type.
* Remove the extra local from local_internals variable names
* Add an extra explanatory comment to local_internals
* Fix a typo in the code
* Raise codec errors when casting to std::string
Allow the codec's exception to be raised instead of RuntimeError when
casting from py::str to std::string.
PY2 allows ucs surrogates in UTF-8 conversion
Signed-off-by: Shane Loretz <sloretz@openrobotics.org>
Signed-off-by: Shane Loretz <sloretz@osrfoundation.org>
* Attempt to fix py2 error
* Revert all unicode literals
* Fixed
Co-authored-by: Aaron Gokaslan <skylion.aaron@gmail.com>
* Manually moving `// NOLINT` comments so that clang-format does not move them to the wrong places.
* Manually reformatting comments related to `static_assert`s so that clang-format does not need two passes.
* Empty lines between #includes, to prevent clang-format from shuffling the order and thereby confusing MSVC 2015.
* git diff -U0 --no-color HEAD^ | python3 $HOME/clone/llvm-project/clang/tools/clang-format/clang-format-diff.py -p1 -style=file -i
For single-file extensions, a convenient pattern offered by cython
is to place the source files directly in the python source tree
(`foo/__init__.py`, `foo/ext.pyx`), deriving the package names from
their filesystem location. Adapt this pattern for pybind11, using an
`intree_extensions` helper, which should be thought of as the moral
equivalent to `cythonize`.
Differences with cythonize: I chose not to include globbing support
(`intree_extensions(glob.glob("**/*.cpp"))` seems sufficient), nor to
provide extension-customization kwargs (directly setting the attributes
on the resulting Pybind11Extension objects seems sufficient).
We could choose to have `intree_extension` (singular instead) and make
users write `[*map(intree_extension, glob.glob("**/*.cpp"))]`; no strong
opinion here.
Co-authored-by: Aaron Gokaslan <skylion.aaron@gmail.com>
* Adding test_return_vector_bool_raw_ptr to test_stl.py.
* First attempt to make the documentation more accurate, but not trying to be comprehensive, to not bloat the reference table with too many details.
* Fixing minor oversights.
* Applying reviewer suggestion.
* Adding iostream.h thread-safety documentation.
* Restoring `TestThread` code with added `std::lock_guard<std::mutex>`.
* Updating new comments to reflect new information.
* Fixing up `git rebase -X theirs` accidents.
* * Rollback of PR #3068.
* Using latest pytest main branch for 3.9 and 3.10.
* WORKAROUND_ENABLING_ROLLBACK_OF_PR3068 in test_class_sh_trampoline_shared_from_this.py
First experiment combining two potential fixes: latest pytest, workaround.
If this succeeds the next step will be to try only latest pytest without the workaround.
Note: the workaround is known to resolve the MSAN error reported under
https://github.com/pybind/pybind11/pull/3068#issuecomment-877658470
* WORKAROUND_ENABLING_ROLLBACK_OF_PR3068 = False
* Narrowing down WORKAROUND_ENABLING_ROLLBACK_OF_PR3068 to Python 3.9
* Splitting out pybind11/stl/filesystem.h.
To solve breakages like: https://github.com/deepmind/open_spiel/runs/2999582108
Mostly following the suggestion here: https://github.com/pybind/pybind11/pull/2730#issuecomment-750507575
Except using pybind11/stl/filesystem.h instead of pybind11/stlfs.h, as decided via chat.
stl.h restored to the exact state before merging PR #2730 via:
```
git checkout 733f8de24f stl.h
```
* Properly including new stl subdirectory in pip wheel config.
This now passes interactively:
```
pytest tests/extra_python_package/
```
* iwyu cleanup.
iwyuh.py -c -std=c++17 -DPYBIND11_TEST_BOOST -Ipybind11/include -I/usr/include/python3.9 -I/usr/include/eigen3 include/pybind11/stl/filesystem.h
* Adding PYBIND11_HAS_FILESYSTEM_IS_OPTIONAL.
* Eliminating else after return.
* Unify cast_error message thrown by [simple|unpacking]_collector
simple_collector and unpacking_collector throw different error messages
when the casting of an argument failed: While the former mentions make_tuple(),
the latter emphasises the call argument (and its name/position).
* Consolidating "Unable to convert call argument" error reporting code to guarantee uniformity.
Co-authored-by: Ralf W. Grosse-Kunstleve <rwgk@google.com>
* pickle setstate: setattr __dict__ only if not empty, to not force use of py::dynamic_attr() unnecessarily.
* Adding unit test.
* Clang 3.6 & 3.7 compatibility.
* PyPy compatibility.
* Minor iwyu fix, additional comment.
* Addressing reviewer requests.
* Applying clang-tidy suggested fixes.
* Adding check_dynamic_cast_SimpleCppDerived, related to issue #3062.
Now also passes the open_spiel iterated_prisoners_dilemma_test ASAN clean, in addition to all pybind11 and PyCLIF unit tests.
The problem was that calling `std::shared_ptr<void>::reset()` with a `void` pointer cannot possibly update the `shared_from_this` `weak_ptr`.
The solution is to store a `shd_ptr_reset` function pointer in `guarded_deleter` (similar in idea to the stored function pointer for calling `delete`).
This commit still includes all debugging code, i.e. is "dirty". The code will be cleaned up after the GitHub CI is fully successful.
