* Ensure ThreadPool is closed in setup_helpers
The ParallelCompile setup helper using a ThreadPool to enable its
parallelism. It does not properly close the pool when it is done with
it.
This can lead to a "Exception ignored in: <function Pool.__del__..."
message with traceback being printed at shutdown.
Use pool.terminate() instead of context manager for Python 2.7
compatibility
* Add note to remove code that supports Python 2
Co-authored-by: Bobby Impollonia <bobby@k13capital.com>
* fix: vs2022 compilation, issue #3477
* silence warning for python 2.7
* disable warning around mbstowcs call
* move disable warning code closer to call site
* turn on vs2022 ci test
* ci: don't run helpers on Windows 2022 & Python 3.5
* limit workaround for stdlib shipped with vs2022 or later
* fix for: limit workaround for stdlib shipped with vs2022 or later
* fix 2 for: limit workaround for stdlib shipped with vs2022 or later
* comment
* ci: add a Windows 2019 run
* ci: add Python 2.7 check too
Co-authored-by: Henry Schreiner <HenrySchreinerIII@gmail.com>
* Expand string_view support to str, bytes, memoryview
1. Allows constructing a str or bytes implicitly from a string_view;
this is essentially a small shortcut allowing a caller to write
`py::bytes{sv}` rather than `py::bytes{sv.data(), sv.size()}`.
2. Allows implicit conversion *to* string_view from py::bytes -- this
saves a fair bit more as currently there is no simple way to get such
a view of the bytes without copying it (or resorting to Python API
calls).
(This is not done for `str` because when the str contains unicode we
have to allocate to a temporary and so there might not be some string
data we can properly view without owning.)
3. Allows `memoryview::from_memory` to accept a string_view. As with
the other from_memory calls, it's entirely your responsibility to
keep it alive.
This also required moving the string_view availability detection into
detail/common.h because this PR needs it in pytypes.h, which is higher
up the include chain than cast.h where it was being detected currently.
* Move string_view include to pytypes.h
* CI-testing a fix for the "ambiguous conversion" issue.
This change is known to fix the `tensorflow::tstring` issue reported under https://github.com/pybind/pybind11/pull/3521#issuecomment-985100965
TODO: Minimal reproducer for the `tensorflow::tstring` issue.
* Make clang-tidy happy (hopefully).
* Adding minimal reproducer for the `tensorflow::tstring` issue.
Error without the enable_if trick:
```
/usr/local/google/home/rwgk/forked/pybind11/tests/test_builtin_casters.cpp:169:16: error: ambiguous conversion for functional-style cast from 'TypeWithBothOperatorStringAndStringView' to 'py::bytes'
return py::bytes(TypeWithBothOperatorStringAndStringView());
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/usr/local/google/home/rwgk/forked/pybind11/include/pybind11/detail/../pytypes.h:1174:5: note: candidate constructor
bytes(const std::string &s) : bytes(s.data(), s.size()) { }
^
/usr/local/google/home/rwgk/forked/pybind11/include/pybind11/detail/../pytypes.h:1191:5: note: candidate constructor
bytes(std::string_view s) : bytes(s.data(), s.size()) { }
^
```
* Adding missing NOLINTNEXTLINE
* Also apply ambiguous conversion workaround to str()
Co-authored-by: Ralf W. Grosse-Kunstleve <rwgk@google.com>
* Replace usage of deprecated Eigen class
Eigen::MappedSparseMatrix has been deprecated since Eigen 3.3 from 2016. Use the equivalent modern syntax Eigen::Map<Eigen::SparseMatrix<...>>.
* Update eigen.h
* Update eigen.h
* fix compiler warning: deprecated implicit copy constructor
* take care of the bug http://eigen.tuxfamily.org/bz/show_bug.cgi?id=747
* add parenthesis for better reading
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* Update include/pybind11/eigen.h
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Henry Schreiner <HenrySchreinerIII@gmail.com>
* Fix py::kw_only when used before the first arg of a method
The implicit space for the `self` argument isn't added until we hit the
first argument, but this wasn't being done for kw_only or pos_only, and
so a kw_only before the first argument would break.
This fixes it by properly checking whether we need to add the self arg.
(The pos_only issue here was extremely mild -- you didn't get the `/` in
the docstring, but AFAICT it has no other effect since there are no
meaningful arguments before it anyway).
* Style changes
- rename check_have_self_arg -> append_self_arg_if_needed
- move the argument name inline comments before the args instead of
after
* ci: support development releases of Python
* fix: better PyPy support
* fix: patch over a few more pypy issues
* Try to patch
* Properly follow pep667
* Fix typo
* Whoops, 667 not in yet
* For testing
* More testing
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* Try to backport
* Try to simplify fix
* Nail down the fix
* Try pypy workaround
* Typo
* one last typo
* Replacing 0x03110000 with 0x030B0000
* Add TODO. Drop PyPy
* Fix typo
* Revert catch upgrade
* fix: minor cleanup, try pypy again
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>
Co-authored-by: Ralf W. Grosse-Kunstleve <rwgk@google.com>
* override: Fix wrong caching of the overrides
There was a problem when the python type, which was stored in override
cache for C++ functions, was destroyed and the record wasn't removed from the
override cache. Therefor, dangling pointer was stored there. Then when the
memory was reused and new type was allocated at the given address and the
method with the same name (as previously stored in the cache) was actually
overridden in python, it would wrongly find it in the override cache for C++
functions and therefor override from python wouldn't be called.
The fix is to erase the type from the override cache when the type is destroyed.
* test: Pass by const ref instead of by value (clang-tidy)
* test: Rename classes and move to different files
Rename the classes and files so they're no too generic. Also, better place to
test the stuff is in test_virtual_functions.cpp/.py as we're basically testing
the virtual functions/trampolines.
* Add TODO for erasure code
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* docs: rework CI a bit, more modern skipping
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
On Unix, setuptools prepends $CFLAGS and $CPPFLAGS to the compiler flags
(they always come before extra_compile_args and anything else; see
distutils.sysconfig.customize_compiler). In practice, the environment
variables are useful e.g. to quickly generate a debug build (e.g. by
setting CFLAGS=-g), but Pybind11Extension currently unconditionally
overwrites this with -g0.
Instead, check the environment variables and only insert -g0 if not
overridden by them.
* fix: add missing std::forward calls
Two of the four cpp_function overloads are missing std::forward calls, which seems like a simple oversight.
* add test for https://github.com/pybind/pybind11/pull/3443
* add py tests
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* fix test
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* Simply has_kw_only_args handling
This simplifies tracking the number of kw-only args by instead tracking
the number of positional arguments (which is really what we care about
everywhere this is used).
* Allow keyword-only arguments to follow py::args
This removes the constraint that py::args has to be last (or
second-last, with py::kwargs) and instead makes py::args imply
py::kw_only for any remaining arguments, allowing you to bind a function
that works the same way as a Python function such as:
def f(a, *args, b):
return a * b + sum(args)
f(10, 1, 2, 3, b=20) # == 206
With this change, you can bind such a function using:
m.def("f", [](int a, py::args args, int b) { /* ... */ },
"a"_a, "b"_a);
Or, to be more explicit about the keyword-only arguments:
m.def("g", [](int a, py::args args, int b) { /* ... */ },
"a"_a, py::kw_only{}, "b"_a);
(The only difference between the two is that the latter will fail at
binding time if the `kw_only{}` doesn't match the `py::args` position).
This doesn't affect backwards compatibility at all because, currently,
you can't have a py::args anywhere except the end/2nd-last.
* Take args/kwargs by const lvalue ref
Co-authored-by: Henry Schreiner <HenrySchreinerIII@gmail.com>
Co-authored-by: Henry Schreiner <HenrySchreinerIII@gmail.com>
* docs: changelog update for 2.8.1
* chore: add one more entry
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* ci: support Python 3.11-dev
Also update 3.10 to final, better PyPy usage
* fix: use PyFrame_GetCode on Python 3.9+
* ci: some bitiness of pypy not supported on win
* chore: update CMake support to 3.22rc1 to quiet warning
* fix: use dev version of py to fix Py 3.11
* tests: print proper Eigen version
* ci: include pypy2, not sure why
* ci: avoid running on Python 3.11 for now
* ci: fix runs
* ci: simpler PyPy usage, drop unmaintained scipy + pypy index
* ci: only binary numpy, wait on pypy 3.8
* refactor: address review
* fix: the types for return_value_policy_override in optional_caster
`return_value_policy_override` was not being applied correctly in
`optional_caster` in two ways:
- The `is_lvalue_reference` condition referenced `T`, which was the
`optional<T>` type parameter from the class, when it should have used `T_`,
which was the parameter to the `cast` function. `T_` can potentially be a
reference type, but `T` will never be.
- The type parameter passed to `return_value_policy_override` should be
`T::value_type`, not `T`. This matches the way that the other STL container
type casters work.
The result of these issues was that a method/property definition which used a
`reference` or `reference_internal` return value policy would create a Python
value that's bound by reference to a temporary C++ object, resulting in
undefined behavior. For reasons that I was not able to figure out fully, it
seems like this causes problems when using old versions of `boost::optional`,
but not with recent versions of `boost::optional` or the `libstdc++`
implementation of `std::optional`. The issue (that the override to
`return_value_policy::move` is never being applied) is present for all
implementations, it just seems like that somehow doesn't result in problems for
the some implementation of `optional`. This change includes a regression type
with a custom optional-like type which was able to reproduce the issue.
Part of the issue with using the wrong types may have stemmed from the type
variables `T` and `T_` having very similar names. This also changes the type
variables in `optional_caster` to use slightly more descriptive names, which
also more closely follow the naming convention used by the other STL casters.
Fixes#3330
* Fix clang-tidy complaints
* Add missing NOLINT
* Apply a couple more fixes
* fix: support GCC 4.8
* tests: avoid warning about unknown compiler for compilers missing C++17
* Remove unneeded test module attribute
* Change test enum to have more unique int values
Co-authored-by: Aaron Gokaslan <skylion.aaron@gmail.com>
Co-authored-by: Henry Schreiner <HenrySchreinerIII@gmail.com>
* 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
