* 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 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>
* 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>
* 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: 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>
* 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>
* Removing last remnants of pragma block at the top of pybind11.h, defaulting CUDA, GCC7, GCC8 to PYBIND11_NOINLINE_DISABLED, with the option to define PYBIND11_NOINLINE_FORCED.
* Unique SOSIZE prefix to make it easier to extract the sosizes from the GitHub logs.
* Commenting out PYBIND11_WERROR block, for noinline testing.
* Undoing accidental change.
* `#define PYBIND11_NOINLINE_FORCED`
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* `#define PYBIND11_NOINLINE_DISABLED`
* Going back to default (removing `#define PYBIND11_NOINLINE_DISABLED`).
* `#define PYBIND11_NOINLINE_FORCED`
* Undoing all changes releated to measuring sosizes.
* Rollback of PR #3030 (Working around Centos 8 failure).
* Disabling -Werror for GNU (experiment).
* Commenting out the entire `if(PYBIND11_WERROR)` again (although that is not expected to make a difference, but who knows what I am overlooking).
* Adding `-DCMAKE_BUILD_TYPE=Release`
* Undoing change to tests/CMakeLists.txt (uncommenting `if(PYBIND11_WERROR)` block).
* post `git rebase master -X theirs` fixups.
* Adding measurements to comment for `PYBIND11_NOINLINE_FORCED`.
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* Removing pragma for GCC -Wattributes, fixing forward declarations.
* Introducing PYBIND11_NOINLINE_FWD to deal with CUDA, GCC7, GCC8.
* Updating PYBIND11_NOINLINE_DCL in Doxyfile.
* Trying noinline, noinline for {CUDA, GCC7, GCC8}
* Trying noinline, inline for {CUDA, GCC7, GCC8}
* Adding GCC -Wattributes `pragma` in 3 header files.
* Introducing PYBIND11_NOINLINE_GCC_PRAGMA_ATTRIBUTES_NEEDED, used in 9 header files.
* Removing ICC pragma 2196, to see if it is still needed.
* Trying noinline, noinline for ICC
* Trying noinline, inline for ICC
* Restoring ICC pragma 2196, introducing PYBIND11_NOINLINE_FORCED, defined for testing.
* Removing code accidentally left in (was for experimentation only).
* Removing one-time-test define.
* Removing PYBIND11_NOINLINE_FWD macro (after learning that it makes no sense).
* Testing with PYBIND11_NOINLINE_DISABLED. Minor non-functional enhancements.
* Removing #define PYBIND11_NOINLINE_DISABLED (test was successful).
* Removing PYBIND11_NOINLINE_FORCED and enhancing comments for PYBIND11_NOINLINE.
* WIP stripping back
* Making -Wattributes pragma in pybind11 specific to GCC7, GCC8, CUDA.
* Cleanup triggered by work on pragma for GCC -Wunused-but-set-parameter.
* Backing out changes to eigen.h (to be worked on later).
* Adding PYBIND11_WORKAROUND_INCORRECT_GCC_UNUSED_BUT_SET_PARAMETER in type_caster_base.h (apparently needed only for older GCCs).
* Apparently older compilers need a simpler overload for silence_unused_warnings().
* clang C++11 compatibility: removing constexpr
* Special case for MSVC 2017: `constexpr void` return
* Trying again without the silence_unused_warnings(const int *) overload.
* Separate macros for ALL_GCC, OLD_GCC_UNUSED_BUT_SET_PARAMETER
* Changing to __GNUC__ <= 2 (turning off)
* Refined condition for PYBIND11_WORKAROUND_INCORRECT_OLD_GCC_UNUSED_BUT_SET_PARAMETER.
* Quick experiment trying out suggestion by @henryiii
* Introducing macro: PYBIND11_INT_ARRAY_WORKING_AROUND_MSVC_CLANG_ISSUES
* Trying henryiii@ (void) expander idea.
* fix: apply simpler expression with fewer workarounds
* Purging new-but-already-obsoleted macro, made possible by @henryiii's commit.
* Renaming `ALL_GCC` macro back to just `GCC` (because there is no `OLD` anymore, luckily).
* [actions skip] Adding "All GCC versions" to comment, to be clear about it.
Co-authored-by: Henry Schreiner <henryschreineriii@gmail.com>
* Removing pragma for 4127 (to see what is still broken with the latest code).
* Using new constexpr_bool() to suppress warning C4127.
* One missed case, Python 2 only.
* PYBIND11_SILENCE_MSVC_C4127 (more similar to the approach for C4100).
* Moving pragma for C4505 from pybind11.h to existing list in detail/common.h.
* Removing 4 existing suppressions to 1. see what is still needed and 2. capture the MSVC messages.
* It turns out none of the 4 pragmas are needed anymore.
* Moving pragma for ignoring -Wnoexcept-type to the one location where it is needed.
* Trying a second location.
* The previous commit worked (GitHub Actions green), but see the added comment about the dicy nature of -Wnoexcept-type ("if and only if").
* Applying reviewer suggestion.
* Removing pragma for 4100 (to see what is still broken with the latest code).
* Adding --keep-going
* Revert "Adding --keep-going"
This reverts commit 1c844c6ffd07a6111b644811e7e3b0a50b9d44bb.
* Introducing PYBIND11_WORKAROUND_INCORRECT_MSVC_C4100.
* _MSC_VER <= 1916
* Replacing simple variadic function with variadic template (attempt to resolve MSVC 2017 failures).
* Preserving existing comment (moved from pybind11.h to detail/common.h).
* Adding blank lines for readability.
* 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
* Adding PYBIND11_COMPAT_BOOL_CAST to appease MSVC 2015 warning C4800.
* Replacing PYBIND11_COMPAT_BOOL_CAST with simpler != 0
* Extra parentheses (almost all compilers failed without these).
* Removing MSVC C4996 from pragma block at the top of pybind11.h
* localtime_thread_safe, PYBIND11_COMPAT_STRDUP
* Adding #include <ctime> (attempt to fix MSVC 2015, 2017 errors).
* 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
* Enable -Wstrict-aliasing warning
* Narrow down the scope of -Wstrict-aliasing
* Go home, MSVC, you're drunk
* Make sure "pragma GCC" is not executed on ICC
Co-authored-by: Yannick Jadoul <yannick.jadoul@belgacom.net>
* When determining if a shared_ptr already exists, use a test on the weak_ptr instead of a try/catch block.
* When determining if a shared_ptr already exists, use a test on the weak_ptr instead of a try/catch block.
* weak_from_this is only available in C++17 and later
* Switch to use feature flag instead of C++ version flag.
* Add Microsoft-specific check.
* Avoid undefined preprocessor macro warning treated as error.
* Simplify shared_from_this in init_holder
* Include <version> in detail/common.h (~stolen~ borrowed from @bstaletic's #2816)
* Move class_::get_shared_from_this to detail::try_get_shared_from_this
* Simplify try_get_shared_from_this by using weak_ptr::lock()
Co-authored-by: Yannick Jadoul <yannick.jadoul@belgacom.net>
* Plug leaking function_record objects when exceptions are thrown
* Plug leak of strdup'ed strings in function_record
* Some extra comments about the function_record ownership dance
* Clean up the function_record better, in case of exceptions
* Demonstrate some extra function_record leaks
* Change DeleteStrings template argument to free_strings runtime argument in destruct(function_record *)
* Zero-state unique_function_record deleter object
* Clarify rvalue reference to unique_ptr parameter in initialize_generic
* Use push_back with const char * instead of emplace_back
* Fix a clang warning from [-Wshadow-field-in-constructor-modified]
warning: modifying constructor parameter 'flag' that shadows a field of 'set_flag' [-Wshadow-field-in-constructor-modified]
* Change name based on review
* Avoid thread termination in scoped_released
Do not call `PyEval_RestoreThread()` from `~gil_scoped_release()` if python runtime is finalizing, as it will result in thread termination in Python runtime newer than 3.6, as documented in https://docs.python.org/3/c-api/init.html#c.PyEval_RestoreThread
Similarly do not call `PyThreadState_DeleteCurrent` from `~gil_scoped_acquire()` if runtime is finalizing.
Discovered while debugging PyTorch crash using Python-3.9 described in https://github.com/pytorch/pytorch/issues/47776
* Simplify _Py_IsFinalizing() availability check
* Fix typo
* Add version agnostic `detail::finalization_guard()`
* Move `finalization_guard` to detail/common.h
And rename it to `is_finalizing`
* Move `is_finalizing()` back to pybind11.h
* Simplify `is_finalizing()` check
One should follow documentation rather than make any assumptions
* feat: disarm
* docs: fix comment
Co-authored-by: Henry Schreiner <henryschreineriii@gmail.com>
