Merge branch 'master' into stable

2024-11-22 05:05:11 +00:00 · 2021-12-28 11:47:54 -05:00 · 2021-12-28 11:47:54 -05:00 · 9371614764
commit 9371614764
parent acae930123 45f792efdd
57 changed files with 1066 additions and 338 deletions
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@ -23,35 +23,39 @@ jobs:
    strategy:
      fail-fast: false
      matrix:
-        runs-on: [ubuntu-latest, windows-latest, macos-latest]
+        runs-on: [ubuntu-latest, windows-2022, macos-latest]
        python:
        - '2.7'
        - '3.5'
        - '3.6'
        - '3.9'
        - '3.10'
-        # - '3.11-dev'
-        - 'pypy-3.7-v7.3.5'
-        # - 'pypy-3.8'
+        - 'pypy-3.7-v7.3.7'
+        - 'pypy-3.8-v7.3.7'

        # Items in here will either be added to the build matrix (if not
        # present), or add new keys to an existing matrix element if all the
        # existing keys match.
        #
-        # We support three optional keys: args (both build), args1 (first
-        # build), and args2 (second build).
+        # We support an optional key: args, for cmake args
        include:
          # Just add a key
          - runs-on: ubuntu-latest
-            python: 3.6
+            python: '3.6'
            args: >
              -DPYBIND11_FINDPYTHON=ON
+              -DCMAKE_CXX_FLAGS="-D_=1"
          - runs-on: windows-latest
-            python: 3.6
+            python: '3.6'
            args: >
              -DPYBIND11_FINDPYTHON=ON
          - runs-on: macos-latest
-            python: pypy-2.7
+            python: 'pypy-2.7'
+          # Inject a couple Windows 2019 runs
+          - runs-on: windows-2019
+            python: '3.9'
+          - runs-on: windows-2019
+            python: '2.7'

    name: "🐍 ${{ matrix.python }} • ${{ matrix.runs-on }} • x64 ${{ matrix.args }}"
    runs-on: ${{ matrix.runs-on }}
@ -65,7 +69,8 @@ jobs:
        python-version: ${{ matrix.python }}

    - name: Setup Boost (Linux)
-      if: runner.os == 'Linux'
+      # Can't use boost + define _
+      if: runner.os == 'Linux' && matrix.python != '3.6'
      run: sudo apt-get install libboost-dev

    - name: Setup Boost (macOS)
@ -122,7 +127,7 @@ jobs:
      run: git clean -fdx

    # Second build - C++17 mode and in a build directory
-    - name: Configure ${{ matrix.args2 }}
+    - name: Configure C++17
      run: >
        cmake -S . -B build2
        -DPYBIND11_WERROR=ON
@ -130,7 +135,6 @@ jobs:
        -DDOWNLOAD_EIGEN=ON
        -DCMAKE_CXX_STANDARD=17
        ${{ matrix.args }}
-        ${{ matrix.args2 }}

    - name: Build
      run: cmake --build build2 -j 2
@ -184,6 +188,7 @@ jobs:
    # setuptools
    - name: Setuptools helpers test
      run: pytest tests/extra_setuptools
+      if: "!(matrix.python == '3.5' && matrix.runs-on == 'windows-2022')"


  deadsnakes:
@ -244,6 +249,8 @@ jobs:
        python -m pip install -r tests/requirements.txt

    - name: Configure
+      env:
+        SETUPTOOLS_USE_DISTUTILS: stdlib
      run: >
        cmake -S . -B build
        -DCMAKE_BUILD_TYPE=Debug
@ -552,37 +559,37 @@ jobs:
        set +e; source /opt/intel/oneapi/setvars.sh; set -e
        cmake --build build-11 --target test_cmake_build

-    - name: Configure C++17
+    - name: Configure C++14
      run: |
        set +e; source /opt/intel/oneapi/setvars.sh; set -e
-        cmake -S . -B build-17     \
+        cmake -S . -B build-14     \
        -DPYBIND11_WERROR=ON    \
        -DDOWNLOAD_CATCH=ON     \
        -DDOWNLOAD_EIGEN=OFF    \
-        -DCMAKE_CXX_STANDARD=17             \
+        -DCMAKE_CXX_STANDARD=14             \
        -DCMAKE_CXX_COMPILER=$(which icpc)  \
        -DPYTHON_EXECUTABLE=$(python3 -c "import sys; print(sys.executable)")

-    - name: Build C++17
+    - name: Build C++14
      run: |
        set +e; source /opt/intel/oneapi/setvars.sh; set -e
-        cmake --build build-17 -j 2 -v
+        cmake --build build-14 -j 2 -v

-    - name: Python tests C++17
+    - name: Python tests C++14
      run: |
        set +e; source /opt/intel/oneapi/setvars.sh; set -e
        sudo service apport stop
-        cmake --build build-17 --target check
+        cmake --build build-14 --target check

-    - name: C++ tests C++17
+    - name: C++ tests C++14
      run: |
        set +e; source /opt/intel/oneapi/setvars.sh; set -e
-        cmake --build build-17 --target cpptest
+        cmake --build build-14 --target cpptest

-    - name: Interface test C++17
+    - name: Interface test C++14
      run: |
        set +e; source /opt/intel/oneapi/setvars.sh; set -e
-        cmake --build build-17 --target test_cmake_build
+        cmake --build build-14 --target test_cmake_build


  # Testing on CentOS (manylinux uses a centos base, and this is an easy way
@ -704,14 +711,11 @@ jobs:
    - name: Install Doxygen
      run: sudo apt-get install -y doxygen librsvg2-bin # Changed to rsvg-convert in 20.04

-    - name: Install docs & setup requirements
-      run: python3 -m pip install -r docs/requirements.txt
-
    - name: Build docs
-      run: python3 -m sphinx -W -b html docs docs/.build
+      run: pipx run nox -s docs

    - name: Make SDist
-      run: python3 setup.py sdist
+      run: pipx run nox -s build -- --sdist

    - run: git status --ignored

@ -723,7 +727,7 @@ jobs:
    - name: Compare Dists (headers only)
      working-directory: include
      run: |
-        python3 -m pip install --user -U ../dist/*
+        python3 -m pip install --user -U ../dist/*.tar.gz
        installed=$(python3 -c "import pybind11; print(pybind11.get_include() + '/pybind11')")
        diff -rq $installed ./pybind11

--- a/.github/workflows/upstream.yml
+++ b/.github/workflows/upstream.yml
@ -0,0 +1,112 @@
+
+name: Upstream
+
+on:
+  workflow_dispatch:
+  pull_request:
+
+concurrency:
+  group: upstream-${{ github.ref }}
+  cancel-in-progress: true
+
+env:
+  PIP_ONLY_BINARY: numpy
+
+jobs:
+  standard:
+    name: "🐍 3.11 dev • ubuntu-latest • x64"
+    runs-on: ubuntu-latest
+    if: "contains(github.event.pull_request.labels.*.name, 'python dev')"
+
+    steps:
+    - uses: actions/checkout@v2
+
+    - name: Setup Python 3.11
+      uses: actions/setup-python@v2
+      with:
+        python-version: "3.11-dev"
+
+    - name: Setup Boost (Linux)
+      if: runner.os == 'Linux'
+      run: sudo apt-get install libboost-dev
+
+    - name: Update CMake
+      uses: jwlawson/actions-setup-cmake@v1.11
+
+    - name: Prepare env
+      run: |
+        python -m pip install -r tests/requirements.txt
+
+    - name: Setup annotations on Linux
+      if: runner.os == 'Linux'
+      run: python -m pip install pytest-github-actions-annotate-failures
+
+    # First build - C++11 mode and inplace
+    - name: Configure C++11
+      run: >
+        cmake -S . -B .
+        -DPYBIND11_WERROR=ON
+        -DDOWNLOAD_CATCH=ON
+        -DDOWNLOAD_EIGEN=ON
+        -DCMAKE_CXX_STANDARD=11
+
+    - name: Build C++11
+      run: cmake --build . -j 2
+
+    - name: Python tests C++11
+      run: cmake --build . --target pytest -j 2
+
+    - name: C++11 tests
+      run: cmake --build .  --target cpptest -j 2
+
+    - name: Interface test C++11
+      run: cmake --build . --target test_cmake_build
+
+    - name: Clean directory
+      run: git clean -fdx
+
+    # Second build - C++17 mode and in a build directory
+    - name: Configure C++17
+      run: >
+        cmake -S . -B build2
+        -DPYBIND11_WERROR=ON
+        -DDOWNLOAD_CATCH=ON
+        -DDOWNLOAD_EIGEN=ON
+        -DCMAKE_CXX_STANDARD=17
+        ${{ matrix.args }}
+        ${{ matrix.args2 }}
+
+    - name: Build
+      run: cmake --build build2 -j 2
+
+    - name: Python tests
+      run: cmake --build build2 --target pytest
+
+    - name: C++ tests
+      run: cmake --build build2 --target cpptest
+
+    # Third build - C++17 mode with unstable ABI
+    - name: Configure (unstable ABI)
+      run: >
+        cmake -S . -B build3
+        -DPYBIND11_WERROR=ON
+        -DDOWNLOAD_CATCH=ON
+        -DDOWNLOAD_EIGEN=ON
+        -DCMAKE_CXX_STANDARD=17
+        -DPYBIND11_INTERNALS_VERSION=10000000
+        "-DPYBIND11_TEST_OVERRIDE=test_call_policies.cpp;test_gil_scoped.cpp;test_thread.cpp"
+        ${{ matrix.args }}
+
+    - name: Build (unstable ABI)
+      run: cmake --build build3 -j 2
+
+    - name: Python tests (unstable ABI)
+      run: cmake --build build3 --target pytest
+
+    - name: Interface test
+      run: cmake --build build2 --target test_cmake_build
+
+    # This makes sure the setup_helpers module can build packages using
+    # setuptools
+    - name: Setuptools helpers test
+      run: pytest tests/extra_setuptools
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@ -15,7 +15,7 @@
 repos:
 # Standard hooks
 - repo: https://github.com/pre-commit/pre-commit-hooks
-  rev: v4.0.1
+  rev: v4.1.0
  hooks:
  - id: check-added-large-files
  - id: check-case-conflict
@ -33,27 +33,27 @@ repos:
    exclude: ^noxfile.py$

 - repo: https://github.com/asottile/pyupgrade
-  rev: v2.29.0
+  rev: v2.29.1
  hooks:
  - id: pyupgrade

 - repo: https://github.com/PyCQA/isort
-  rev: 5.9.3
+  rev: 5.10.1
  hooks:
  - id: isort

 # Black, the code formatter, natively supports pre-commit
 - repo: https://github.com/psf/black
-  rev: 21.9b0 # Keep in sync with blacken-docs
+  rev: 21.12b0 # Keep in sync with blacken-docs
  hooks:
  - id: black

 - repo: https://github.com/asottile/blacken-docs
-  rev: v1.11.0
+  rev: v1.12.0
  hooks:
  - id: blacken-docs
    additional_dependencies:
-    - black==21.9b0 # keep in sync with black hook
+    - black==21.12b0 # keep in sync with black hook

 # Changes tabs to spaces
 - repo: https://github.com/Lucas-C/pre-commit-hooks
@ -61,6 +61,12 @@ repos:
  hooks:
  - id: remove-tabs

+# Autoremoves unused imports
+- repo: https://github.com/hadialqattan/pycln
+  rev: v1.1.0
+  hooks:
+  - id: pycln
+
 - repo: https://github.com/pre-commit/pygrep-hooks
  rev: v1.9.0
  hooks:
@ -98,7 +104,7 @@ repos:

 # Check static types with mypy
 - repo: https://github.com/pre-commit/mirrors-mypy
-  rev: v0.910-1
+  rev: v0.930
  hooks:
  - id: mypy
    # Running per-file misbehaves a bit, so just run on all files, it's fast
@ -122,7 +128,7 @@ repos:
    args: ["-L", "nd,ot,thist"]

 - repo: https://github.com/shellcheck-py/shellcheck-py
-  rev: v0.7.2.1
+  rev: v0.8.0.3
  hooks:
  - id: shellcheck

--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@ -16,6 +16,11 @@ else()
  cmake_policy(VERSION 3.22)
 endif()

+# Avoid infinite recursion if tests include this as a subdirectory
+if(DEFINED PYBIND11_MASTER_PROJECT)
+  return()
+endif()
+
 # Extract project version from source
 file(STRINGS "${CMAKE_CURRENT_SOURCE_DIR}/include/pybind11/detail/common.h"
     pybind11_version_defines REGEX "#define PYBIND11_VERSION_(MAJOR|MINOR|PATCH) ")
@ -45,13 +50,8 @@ if(NOT pybind11_FIND_QUIETLY)
  message(STATUS "pybind11 v${pybind11_VERSION} ${pybind11_VERSION_TYPE}")
 endif()

-# Avoid infinite recursion if tests include this as a subdirectory
-if(DEFINED PYBIND11_MASTER_PROJECT)
-  set(PYBIND11_TEST OFF)
-endif()
-
 # Check if pybind11 is being used directly or via add_subdirectory
-if(CMAKE_SOURCE_DIR STREQUAL PROJECT_SOURCE_DIR AND NOT DEFINED PYBIND11_MASTER_PROJECT)
+if(CMAKE_SOURCE_DIR STREQUAL PROJECT_SOURCE_DIR)
  ### Warn if not an out-of-source builds
  if(CMAKE_CURRENT_SOURCE_DIR STREQUAL CMAKE_CURRENT_BINARY_DIR)
    set(lines
@ -80,6 +80,8 @@ if(CMAKE_SOURCE_DIR STREQUAL PROJECT_SOURCE_DIR AND NOT DEFINED PYBIND11_MASTER_
  endif()

  set(pybind11_system "")
+
+  set_property(GLOBAL PROPERTY USE_FOLDERS ON)
 else()
  set(PYBIND11_MASTER_PROJECT OFF)
  set(pybind11_system SYSTEM)
--- a/README.rst
+++ b/README.rst
@ -106,7 +106,7 @@ goodies:
 - Binaries are generally smaller by a factor of at least 2 compared to
  equivalent bindings generated by Boost.Python. A recent pybind11
  conversion of PyRosetta, an enormous Boost.Python binding project,
-  `reported <http://graylab.jhu.edu/RosettaCon2016/PyRosetta-4.pdf>`_
+  `reported <https://graylab.jhu.edu/Sergey/2016.RosettaCon/PyRosetta-4.pdf>`_
  a binary size reduction of **5.4x** and compile time reduction by
  **5.8x**.

--- a/docs/Doxyfile
+++ b/docs/Doxyfile
@ -18,6 +18,5 @@ ALIASES               += "endrst=\endverbatim"
 QUIET                  = YES
 WARNINGS               = YES
 WARN_IF_UNDOCUMENTED   = NO
-PREDEFINED             = DOXYGEN_SHOULD_SKIP_THIS \
-                         PY_MAJOR_VERSION=3 \
+PREDEFINED             = PY_MAJOR_VERSION=3 \
                         PYBIND11_NOINLINE
--- a/docs/advanced/cast/custom.rst
+++ b/docs/advanced/cast/custom.rst
@ -46,7 +46,7 @@ type is explicitly allowed.
             * function signatures and declares a local variable
             * 'value' of type inty
             */
-            PYBIND11_TYPE_CASTER(inty, _("inty"));
+            PYBIND11_TYPE_CASTER(inty, const_name("inty"));

            /**
             * Conversion part 1 (Python->C++): convert a PyObject into a inty
--- a/docs/advanced/cast/eigen.rst
+++ b/docs/advanced/cast/eigen.rst
@ -52,7 +52,7 @@ can be mapped *and* if the numpy array is writeable (that is
 the passed variable will be transparently carried out directly on the
 ``numpy.ndarray``.

-This means you can can write code such as the following and have it work as
+This means you can write code such as the following and have it work as
 expected:

 .. code-block:: cpp
--- a/docs/advanced/cast/overview.rst
+++ b/docs/advanced/cast/overview.rst
@ -75,96 +75,96 @@ The following basic data types are supported out of the box (some may require
 an additional extension header to be included). To pass other data structures
 as arguments and return values, refer to the section on binding :ref:`classes`.

-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 |  Data type                         |  Description              | Header file                       |
-+====================================+===========================+===============================+
+====================================+===========================+===================================+
 | ``int8_t``, ``uint8_t``            | 8-bit integers            | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``int16_t``, ``uint16_t``          | 16-bit integers           | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``int32_t``, ``uint32_t``          | 32-bit integers           | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``int64_t``, ``uint64_t``          | 64-bit integers           | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``ssize_t``, ``size_t``            | Platform-dependent size   | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``float``, ``double``              | Floating point types      | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``bool``                           | Two-state Boolean type    | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``char``                           | Character literal         | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``char16_t``                       | UTF-16 character literal  | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``char32_t``                       | UTF-32 character literal  | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``wchar_t``                        | Wide character literal    | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``const char *``                   | UTF-8 string literal      | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``const char16_t *``               | UTF-16 string literal     | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``const char32_t *``               | UTF-32 string literal     | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``const wchar_t *``                | Wide string literal       | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::string``                    | STL dynamic UTF-8 string  | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::u16string``                 | STL dynamic UTF-16 string | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::u32string``                 | STL dynamic UTF-32 string | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::wstring``                   | STL dynamic wide string   | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::string_view``,              | STL C++17 string views    | :file:`pybind11/pybind11.h`       |
 | ``std::u16string_view``, etc.      |                           |                                   |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::pair<T1, T2>``              | Pair of two custom types  | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::tuple<...>``                | Arbitrary tuple of types  | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::reference_wrapper<...>``    | Reference type wrapper    | :file:`pybind11/pybind11.h`       |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::complex<T>``                | Complex numbers           | :file:`pybind11/complex.h`        |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::array<T, Size>``            | STL static array          | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::vector<T>``                 | STL dynamic array         | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::deque<T>``                  | STL double-ended queue    | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::valarray<T>``               | STL value array           | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::list<T>``                   | STL linked list           | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::map<T1, T2>``               | STL ordered map           | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::unordered_map<T1, T2>``     | STL unordered map         | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::set<T>``                    | STL ordered set           | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::unordered_set<T>``          | STL unordered set         | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::optional<T>``               | STL optional type (C++17) | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::experimental::optional<T>`` | STL optional type (exp.)  | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::variant<...>``              | Type-safe union (C++17)   | :file:`pybind11/stl.h`            |
-+------------------------------------+---------------------------+-------------------------------+
-| ``std::filesystem::path<T>``       | STL path (C++17) [#]_     | :file:`pybind11/stl.h`        |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
+| ``std::filesystem::path<T>``       | STL path (C++17) [#]_     | :file:`pybind11/stl/filesystem.h` |
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::function<...>``             | STL polymorphic function  | :file:`pybind11/functional.h`     |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::chrono::duration<...>``     | STL time duration         | :file:`pybind11/chrono.h`         |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``std::chrono::time_point<...>``   | STL date/time             | :file:`pybind11/chrono.h`         |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``Eigen::Matrix<...>``             | Eigen: dense matrix       | :file:`pybind11/eigen.h`          |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``Eigen::Map<...>``                | Eigen: mapped memory      | :file:`pybind11/eigen.h`          |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+
 | ``Eigen::SparseMatrix<...>``       | Eigen: sparse matrix      | :file:`pybind11/eigen.h`          |
-+------------------------------------+---------------------------+-------------------------------+
+------------------------------------+---------------------------+-----------------------------------+

 .. [#] ``std::filesystem::path`` is converted to ``pathlib.Path`` and
   ``os.PathLike`` is converted to ``std::filesystem::path``, but this requires
--- a/docs/advanced/exceptions.rst
+++ b/docs/advanced/exceptions.rst
@ -64,7 +64,7 @@ at its exception handler.
 +--------------------------------------+--------------------------------------+

 Exception translation is not bidirectional. That is, *catching* the C++
-exceptions defined above above will not trap exceptions that originate from
+exceptions defined above will not trap exceptions that originate from
 Python. For that, catch :class:`pybind11::error_already_set`. See :ref:`below
 <handling_python_exceptions_cpp>` for further details.

--- a/docs/advanced/functions.rst
+++ b/docs/advanced/functions.rst
@ -306,8 +306,9 @@ The class ``py::args`` derives from ``py::tuple`` and ``py::kwargs`` derives
 from ``py::dict``.

 You may also use just one or the other, and may combine these with other
-arguments as long as the ``py::args`` and ``py::kwargs`` arguments are the last
-arguments accepted by the function.
+arguments.  Note, however, that ``py::kwargs`` must always be the last argument
+of the function, and ``py::args`` implies that any further arguments are
+keyword-only (see :ref:`keyword_only_arguments`).

 Please refer to the other examples for details on how to iterate over these,
 and on how to cast their entries into C++ objects. A demonstration is also
@ -366,6 +367,8 @@ like so:
    py::class_<MyClass>("MyClass")
        .def("myFunction", py::arg("arg") = static_cast<SomeType *>(nullptr));

+.. _keyword_only_arguments:
+
 Keyword-only arguments
 ======================

@ -397,6 +400,15 @@ feature does *not* require Python 3 to work.

 .. versionadded:: 2.6

+As of pybind11 2.9, a ``py::args`` argument implies that any following arguments
+are keyword-only, as if ``py::kw_only()`` had been specified in the same
+relative location of the argument list as the ``py::args`` argument.  The
+``py::kw_only()`` may be included to be explicit about this, but is not
+required.  (Prior to 2.9 ``py::args`` may only occur at the end of the argument
+list, or immediately before a ``py::kwargs`` argument at the end).
+
+.. versionadded:: 2.9
+
 Positional-only arguments
 =========================

--- a/docs/basics.rst
+++ b/docs/basics.rst
@ -109,7 +109,7 @@ a file named :file:`example.cpp` with the following contents:
    PYBIND11_MODULE(example, m) {
        m.doc() = "pybind11 example plugin"; // optional module docstring

-        m.def("add", &add, "A function which adds two numbers");
+        m.def("add", &add, "A function that adds two numbers");
    }

 .. [#f1] In practice, implementation and binding code will generally be located
--- a/docs/changelog.rst
+++ b/docs/changelog.rst
@ -7,8 +7,100 @@ Starting with version 1.8.0, pybind11 releases use a `semantic versioning
 <http://semver.org>`_ policy.


-IN DEVELOPMENT
--------------
+Version 2.9.0 (Dec 28, 2021)
+----------------------------
+
+This is the last version to support Python 2.7 and 3.5.
+
+New Features:
+
+* Allow ``py::args`` to be followed by other arguments; the remaining arguments
+  are implicitly keyword-only, as if a ``py::kw_only{}`` annotation had been
+  used.
+  `#3402 <https://github.com/pybind/pybind11/pull/3402>`_
+
+* Add C++ Exception type to throw and catch ``AttributeError``. Useful for
+  defining custom ``__setattr__`` and ``__getattr__`` methods.
+  `#3387 <https://github.com/pybind/pybind11/pull/3387>`_
+
+Changes:
+
+* Make str/bytes/memoryview more interoperable with ``std::string_view``.
+  `#3521 <https://github.com/pybind/pybind11/pull/3521>`_
+
+* Replace ``_`` with ``const_name`` in internals, avoid defining ``pybind::_``
+  if ``_`` defined as macro (common gettext usage)
+  `#3423 <https://github.com/pybind/pybind11/pull/3423>`_
+
+
+Bug fixes:
+
+* Fix a regression in 2.8.0 that caused undefined behavior (typically
+  segfaults) in ``make_key_iterator``/``make_value_iterator`` if dereferencing
+  the iterator returned a temporary value instead of a reference.
+  `#3348 <https://github.com/pybind/pybind11/pull/3348>`_
+
+* Fix a rare warning about extra copy in an Eigen constructor.
+  `#3486 <https://github.com/pybind/pybind11/pull/3486>`_
+
+* Fix caching of the C++ overrides.
+  `#3465 <https://github.com/pybind/pybind11/pull/3465>`_
+
+* Add missing ``std::forward`` calls to some ``cpp_function`` overloads.
+  `#3443 <https://github.com/pybind/pybind11/pull/3443>`_
+
+* Support PyPy 7.3.7 and the PyPy3.8 beta. Test python-3.11 on PRs with the
+  ``python dev`` label.
+  `#3419 <https://github.com/pybind/pybind11/pull/3419>`_
+
+* Fix 2.8.0 regression with MSVC 2017 + C++17 mode + Python 3.
+  `#3407 <https://github.com/pybind/pybind11/pull/3407>`_
+
+* Modernize usage of ``PyCodeObject`` on Python 3.9 (toward supporting Python
+  3.11a1)
+  `#3368 <https://github.com/pybind/pybind11/pull/3368>`_
+
+* A long-standing bug in eigen.h was fixed (originally PR #3343). The bug was
+  unmasked by newly added ``static_assert``'s in the Eigen 3.4.0 release.
+  `#3352 <https://github.com/pybind/pybind11/pull/3352>`_
+
+* Replace usage of deprecated ``Eigen::MappedSparseMatrix`` with
+  ``Eigen::Map<Eigen::SparseMatrix<...>>`` for Eigen 3.3+.
+  `#3499 <https://github.com/pybind/pybind11/pull/3499>`_
+
+* Fixed the potential for dangling references when using properties with
+  ``std::optional`` types.
+  `#3376 <https://github.com/pybind/pybind11/pull/3376>`_
+
+* Tweaks to support Microsoft Visual Studio 2022.
+  `#3497 <https://github.com/pybind/pybind11/pull/3497>`_
+
+Build system improvements:
+
+* Nicer CMake printout and IDE organisation for pybind11's own tests.
+  `#3479 <https://github.com/pybind/pybind11/pull/3479>`_
+
+* CMake: report version type as part of the version string to avoid a spurious
+  space in the package status message.
+  `#3472 <https://github.com/pybind/pybind11/pull/3472>`_
+
+* Support multiple raw inclusion of CMake helper files (Conan.io does this for
+  multi-config generators).
+  `#3420 <https://github.com/pybind/pybind11/pull/3420>`_
+
+* Fix harmless warning on CMake 3.22.
+  `#3368 <https://github.com/pybind/pybind11/pull/3368>`_
+
+* Flags starting with ``-g`` in ``$CFLAGS`` and ``$CPPFLAGS`` are no longer
+  overridden by ``.Pybind11Extension``.
+  `#3436 <https://github.com/pybind/pybind11/pull/3436>`_
+
+* Ensure ThreadPool is closed in ``setup_helpers``.
+  `#3548 <https://github.com/pybind/pybind11/pull/3548>`_
+
+* Avoid LTS on ``mips64`` and ``ppc64le`` (reported broken).
+  `#3557 <https://github.com/pybind/pybind11/pull/3557>`_
+

 v2.8.1 (Oct 27, 2021)
 ---------------------
--- a/docs/release.rst
+++ b/docs/release.rst
@ -22,6 +22,9 @@ the version just below.
 To release a new version of pybind11:
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

+If you don't have nox, you should either use ``pipx run nox`` instead, or use
+``pipx install nox`` or ``brew install nox`` (Unix).
+
 - Update the version number
    - Update ``PYBIND11_VERSION_MAJOR`` etc. in
      ``include/pybind11/detail/common.h``. PATCH should be a simple integer.
@ -51,14 +54,12 @@ To release a new version of pybind11:
  notifications to users watching releases, and also uploads PyPI packages).
  (Note: if you do not use an existing tag, this creates a new lightweight tag
  for you, so you could skip the above step.)
-
    - GUI method: Under `releases <https://github.com/pybind/pybind11/releases>`_
      click "Draft a new release" on the far right, fill in the tag name
      (if you didn't tag above, it will be made here), fill in a release name
      like "Version X.Y.Z", and copy-and-paste the markdown-formatted (!) changelog
      into the description (usually ``cat docs/changelog.rst | pandoc -f rst -t gfm``).
      Check "pre-release" if this is a beta/RC.
-
    - CLI method: with ``gh`` installed, run ``gh release create vX.Y.Z -t "Version X.Y.Z"``
      If this is a pre-release, add ``-p``.

@ -90,9 +91,7 @@ If you need to manually upload releases, you can download the releases from the

 .. code-block:: bash

-    python3 -m pip install build
-    python3 -m build
-    PYBIND11_SDIST_GLOBAL=1 python3 -m build
+    nox -s build
    twine upload dist/*

 This makes SDists and wheels, and the final line uploads them.
--- a/docs/requirements.txt
+++ b/docs/requirements.txt
@ -1,8 +1,5 @@
-breathe==4.26.1
-# docutils 0.17 breaks HTML tags & RTD theme
-# https://github.com/sphinx-doc/sphinx/issues/9001
-docutils==0.16
-sphinx==3.3.1
-sphinx_rtd_theme==0.5.0
-sphinxcontrib-moderncmakedomain==3.17
-sphinxcontrib-svg2pdfconverter==1.1.0
+breathe==4.31.0
+sphinx==3.5.4
+sphinx_rtd_theme==1.0.0
+sphinxcontrib-moderncmakedomain==3.19
+sphinxcontrib-svg2pdfconverter==1.1.1
--- a/docs/upgrade.rst
+++ b/docs/upgrade.rst
@ -17,6 +17,10 @@ v2.9
  converted to using ``py::module_::import("types").attr("SimpleNamespace")``
  instead.

+* The use of ``_`` in custom type casters can now be replaced with the more
+  readable ``const_name`` instead. The old ``_`` shortcut has been retained
+  unless it is being used as a macro (like for gettext).
+

 .. _upgrade-guide-2.7:

--- a/include/pybind11/attr.h
+++ b/include/pybind11/attr.h
@ -174,7 +174,7 @@ struct function_record {
    function_record()
        : is_constructor(false), is_new_style_constructor(false), is_stateless(false),
          is_operator(false), is_method(false), has_args(false),
-          has_kwargs(false), has_kw_only_args(false), prepend(false) { }
+          has_kwargs(false), prepend(false) { }

    /// Function name
    char *name = nullptr; /* why no C++ strings? They generate heavier code.. */
@ -221,17 +221,15 @@ struct function_record {
    /// True if the function has a '**kwargs' argument
    bool has_kwargs : 1;

-    /// True once a 'py::kw_only' is encountered (any following args are keyword-only)
-    bool has_kw_only_args : 1;
-
    /// True if this function is to be inserted at the beginning of the overload resolution chain
    bool prepend : 1;

    /// Number of arguments (including py::args and/or py::kwargs, if present)
    std::uint16_t nargs;

-    /// Number of trailing arguments (counted in `nargs`) that are keyword-only
-    std::uint16_t nargs_kw_only = 0;
+    /// Number of leading positional arguments, which are terminated by a py::args or py::kwargs
+    /// argument or by a py::kw_only annotation.
+    std::uint16_t nargs_pos = 0;

    /// Number of leading arguments (counted in `nargs`) that are positional-only
    std::uint16_t nargs_pos_only = 0;
@ -411,20 +409,23 @@ template <> struct process_attribute<is_new_style_constructor> : process_attribu
    static void init(const is_new_style_constructor &, function_record *r) { r->is_new_style_constructor = true; }
 };

-inline void process_kw_only_arg(const arg &a, function_record *r) {
-    if (!a.name || a.name[0] == '\0')
-        pybind11_fail("arg(): cannot specify an unnamed argument after an kw_only() annotation");
-    ++r->nargs_kw_only;
+inline void check_kw_only_arg(const arg &a, function_record *r) {
+    if (r->args.size() > r->nargs_pos && (!a.name || a.name[0] == '\0'))
+        pybind11_fail("arg(): cannot specify an unnamed argument after a kw_only() annotation or args() argument");
+}
+
+inline void append_self_arg_if_needed(function_record *r) {
+    if (r->is_method && r->args.empty())
+        r->args.emplace_back("self", nullptr, handle(), /*convert=*/ true, /*none=*/ false);
 }

 /// Process a keyword argument attribute (*without* a default value)
 template <> struct process_attribute<arg> : process_attribute_default<arg> {
    static void init(const arg &a, function_record *r) {
-        if (r->is_method && r->args.empty())
-            r->args.emplace_back("self", nullptr, handle(), true /*convert*/, false /*none not allowed*/);
+        append_self_arg_if_needed(r);
        r->args.emplace_back(a.name, nullptr, handle(), !a.flag_noconvert, a.flag_none);

-        if (r->has_kw_only_args) process_kw_only_arg(a, r);
+        check_kw_only_arg(a, r);
    }
 };

@ -432,7 +433,7 @@ template <> struct process_attribute<arg> : process_attribute_default<arg> {
 template <> struct process_attribute<arg_v> : process_attribute_default<arg_v> {
    static void init(const arg_v &a, function_record *r) {
        if (r->is_method && r->args.empty())
-            r->args.emplace_back("self", nullptr /*descr*/, handle() /*parent*/, true /*convert*/, false /*none not allowed*/);
+            r->args.emplace_back("self", /*descr=*/ nullptr, /*parent=*/ handle(), /*convert=*/ true, /*none=*/ false);

        if (!a.value) {
 #if !defined(NDEBUG)
@ -457,21 +458,28 @@ template <> struct process_attribute<arg_v> : process_attribute_default<arg_v> {
        }
        r->args.emplace_back(a.name, a.descr, a.value.inc_ref(), !a.flag_noconvert, a.flag_none);

-        if (r->has_kw_only_args) process_kw_only_arg(a, r);
+        check_kw_only_arg(a, r);
    }
 };

 /// Process a keyword-only-arguments-follow pseudo argument
 template <> struct process_attribute<kw_only> : process_attribute_default<kw_only> {
    static void init(const kw_only &, function_record *r) {
-        r->has_kw_only_args = true;
+        append_self_arg_if_needed(r);
+        if (r->has_args && r->nargs_pos != static_cast<std::uint16_t>(r->args.size()))
+            pybind11_fail("Mismatched args() and kw_only(): they must occur at the same relative argument location (or omit kw_only() entirely)");
+        r->nargs_pos = static_cast<std::uint16_t>(r->args.size());
    }
 };

 /// Process a positional-only-argument maker
 template <> struct process_attribute<pos_only> : process_attribute_default<pos_only> {
    static void init(const pos_only &, function_record *r) {
+        append_self_arg_if_needed(r);
        r->nargs_pos_only = static_cast<std::uint16_t>(r->args.size());
+        if (r->nargs_pos_only > r->nargs_pos)
+            pybind11_fail("pos_only(): cannot follow a py::args() argument");
+            // It also can't follow a kw_only, but a static_assert in pybind11.h checks that
    }
 };

--- a/include/pybind11/cast.h
+++ b/include/pybind11/cast.h
@ -27,23 +27,6 @@
 #include <utility>
 #include <vector>

-#if defined(PYBIND11_CPP17)
-#  if defined(__has_include)
-#    if __has_include(<string_view>)
-#      define PYBIND11_HAS_STRING_VIEW
-#    endif
-#  elif defined(_MSC_VER)
-#    define PYBIND11_HAS_STRING_VIEW
-#  endif
-#endif
-#ifdef PYBIND11_HAS_STRING_VIEW
-#include <string_view>
-#endif
-
-#if defined(__cpp_lib_char8_t) && __cpp_lib_char8_t >= 201811L
-#  define PYBIND11_HAS_U8STRING
-#endif
-
 PYBIND11_NAMESPACE_BEGIN(PYBIND11_NAMESPACE)
 PYBIND11_NAMESPACE_BEGIN(detail)

@ -150,7 +133,8 @@ public:
            return false;
        } else {
            handle src_or_index = src;
-#if PY_VERSION_HEX < 0x03080000
+            // PyPy: 7.3.7's 3.8 does not implement PyLong_*'s __index__ calls.
+#if PY_VERSION_HEX < 0x03080000 || defined(PYPY_VERSION)
            object index;
            if (!PYBIND11_LONG_CHECK(src.ptr())) {  // So: index_check(src.ptr())
                index = reinterpret_steal<object>(PyNumber_Index(src.ptr()));
@ -224,7 +208,7 @@ public:
        return PyLong_FromUnsignedLongLong((unsigned long long) src);
    }

-    PYBIND11_TYPE_CASTER(T, _<std::is_integral<T>::value>("int", "float"));
+    PYBIND11_TYPE_CASTER(T, const_name<std::is_integral<T>::value>("int", "float"));
 };

 template<typename T> struct void_caster {
@ -237,7 +221,7 @@ public:
    static handle cast(T, return_value_policy /* policy */, handle /* parent */) {
        return none().inc_ref();
    }
-    PYBIND11_TYPE_CASTER(T, _("None"));
+    PYBIND11_TYPE_CASTER(T, const_name("None"));
 };

 template <> class type_caster<void_type> : public void_caster<void_type> {};
@ -280,7 +264,7 @@ public:

    template <typename T> using cast_op_type = void*&;
    explicit operator void *&() { return value; }
-    static constexpr auto name = _("capsule");
+    static constexpr auto name = const_name("capsule");
 private:
    void *value = nullptr;
 };
@ -331,7 +315,7 @@ public:
    static handle cast(bool src, return_value_policy /* policy */, handle /* parent */) {
        return handle(src ? Py_True : Py_False).inc_ref();
    }
-    PYBIND11_TYPE_CASTER(bool, _("bool"));
+    PYBIND11_TYPE_CASTER(bool, const_name("bool"));
 };

 // Helper class for UTF-{8,16,32} C++ stl strings:
@ -421,7 +405,7 @@ template <typename StringType, bool IsView = false> struct string_caster {
        return s;
    }

-    PYBIND11_TYPE_CASTER(StringType, _(PYBIND11_STRING_NAME));
+    PYBIND11_TYPE_CASTER(StringType, const_name(PYBIND11_STRING_NAME));

 private:
    static handle decode_utfN(const char *buffer, ssize_t nbytes) {
@ -558,7 +542,7 @@ public:
        return one_char;
    }

-    static constexpr auto name = _(PYBIND11_STRING_NAME);
+    static constexpr auto name = const_name(PYBIND11_STRING_NAME);
    template <typename _T> using cast_op_type = pybind11::detail::cast_op_type<_T>;
 };

@ -595,7 +579,7 @@ public:
        return cast(*src, policy, parent);
    }

-    static constexpr auto name = _("Tuple[") + concat(make_caster<Ts>::name...) + _("]");
+    static constexpr auto name = const_name("Tuple[") + concat(make_caster<Ts>::name...) + const_name("]");

    template <typename T> using cast_op_type = type;

@ -780,14 +764,14 @@ template <typename base, typename holder> struct is_holder_type :
 template <typename base, typename deleter> struct is_holder_type<base, std::unique_ptr<base, deleter>> :
    std::true_type {};

-template <typename T> struct handle_type_name { static constexpr auto name = _<T>(); };
-template <> struct handle_type_name<bytes> { static constexpr auto name = _(PYBIND11_BYTES_NAME); };
-template <> struct handle_type_name<int_> { static constexpr auto name = _("int"); };
-template <> struct handle_type_name<iterable> { static constexpr auto name = _("Iterable"); };
-template <> struct handle_type_name<iterator> { static constexpr auto name = _("Iterator"); };
-template <> struct handle_type_name<none> { static constexpr auto name = _("None"); };
-template <> struct handle_type_name<args> { static constexpr auto name = _("*args"); };
-template <> struct handle_type_name<kwargs> { static constexpr auto name = _("**kwargs"); };
+template <typename T> struct handle_type_name { static constexpr auto name = const_name<T>(); };
+template <> struct handle_type_name<bytes> { static constexpr auto name = const_name(PYBIND11_BYTES_NAME); };
+template <> struct handle_type_name<int_> { static constexpr auto name = const_name("int"); };
+template <> struct handle_type_name<iterable> { static constexpr auto name = const_name("Iterable"); };
+template <> struct handle_type_name<iterator> { static constexpr auto name = const_name("Iterator"); };
+template <> struct handle_type_name<none> { static constexpr auto name = const_name("None"); };
+template <> struct handle_type_name<args> { static constexpr auto name = const_name("*args"); };
+template <> struct handle_type_name<kwargs> { static constexpr auto name = const_name("**kwargs"); };

 template <typename type>
 struct pyobject_caster {
@ -1130,6 +1114,9 @@ constexpr arg operator"" _a(const char *name, size_t) { return arg(name); }

 PYBIND11_NAMESPACE_BEGIN(detail)

+template <typename T> using is_kw_only = std::is_same<intrinsic_t<T>, kw_only>;
+template <typename T> using is_pos_only = std::is_same<intrinsic_t<T>, pos_only>;
+
 // forward declaration (definition in attr.h)
 struct function_record;

@ -1165,17 +1152,18 @@ class argument_loader {

    template <typename Arg> using argument_is_args   = std::is_same<intrinsic_t<Arg>, args>;
    template <typename Arg> using argument_is_kwargs = std::is_same<intrinsic_t<Arg>, kwargs>;
-    // Get args/kwargs argument positions relative to the end of the argument list:
-    static constexpr auto args_pos = constexpr_first<argument_is_args, Args...>() - (int) sizeof...(Args),
-                        kwargs_pos = constexpr_first<argument_is_kwargs, Args...>() - (int) sizeof...(Args);
+    // Get kwargs argument position, or -1 if not present:
+    static constexpr auto kwargs_pos = constexpr_last<argument_is_kwargs, Args...>();

-    static constexpr bool args_kwargs_are_last = kwargs_pos >= - 1 && args_pos >= kwargs_pos - 1;
-
-    static_assert(args_kwargs_are_last, "py::args/py::kwargs are only permitted as the last argument(s) of a function");
+    static_assert(kwargs_pos == -1 || kwargs_pos == (int) sizeof...(Args) - 1, "py::kwargs is only permitted as the last argument of a function");

 public:
-    static constexpr bool has_kwargs = kwargs_pos < 0;
-    static constexpr bool has_args = args_pos < 0;
+    static constexpr bool has_kwargs = kwargs_pos != -1;
+
+    // py::args argument position; -1 if not present.
+    static constexpr int args_pos = constexpr_last<argument_is_args, Args...>();
+
+    static_assert(args_pos == -1 || args_pos == constexpr_first<argument_is_args, Args...>(), "py::args cannot be specified more than once");

    static constexpr auto arg_names = concat(type_descr(make_caster<Args>::name)...);

--- a/include/pybind11/chrono.h
+++ b/include/pybind11/chrono.h
@ -97,7 +97,7 @@ public:
        return PyDelta_FromDSU(dd.count(), ss.count(), us.count());
    }

-    PYBIND11_TYPE_CASTER(type, _("datetime.timedelta"));
+    PYBIND11_TYPE_CASTER(type, const_name("datetime.timedelta"));
 };

 inline std::tm *localtime_thread_safe(const std::time_t *time, std::tm *buf) {
@ -195,7 +195,7 @@ public:
                                          localtime.tm_sec,
                                          us.count());
    }
-    PYBIND11_TYPE_CASTER(type, _("datetime.datetime"));
+    PYBIND11_TYPE_CASTER(type, const_name("datetime.datetime"));
 };

 // Other clocks that are not the system clock are not measured as datetime.datetime objects
--- a/include/pybind11/complex.h
+++ b/include/pybind11/complex.h
@ -59,7 +59,7 @@ public:
        return PyComplex_FromDoubles((double) src.real(), (double) src.imag());
    }

-    PYBIND11_TYPE_CASTER(std::complex<T>, _("complex"));
+    PYBIND11_TYPE_CASTER(std::complex<T>, const_name("complex"));
 };
 PYBIND11_NAMESPACE_END(detail)
 PYBIND11_NAMESPACE_END(PYBIND11_NAMESPACE)
--- a/include/pybind11/detail/common.h
+++ b/include/pybind11/detail/common.h
@ -10,12 +10,12 @@
 #pragma once

 #define PYBIND11_VERSION_MAJOR 2
-#define PYBIND11_VERSION_MINOR 8
-#define PYBIND11_VERSION_PATCH 1
+#define PYBIND11_VERSION_MINOR 9
+#define PYBIND11_VERSION_PATCH 0

 // Similar to Python's convention: https://docs.python.org/3/c-api/apiabiversion.html
 // Additional convention: 0xD = dev
-#define PYBIND11_VERSION_HEX 0x02080100
+#define PYBIND11_VERSION_HEX 0x02090000

 #define PYBIND11_NAMESPACE_BEGIN(name) namespace name {
 #define PYBIND11_NAMESPACE_END(name) }
@ -154,6 +154,14 @@
 // C4505: 'PySlice_GetIndicesEx': unreferenced local function has been removed (PyPy only)
 #  pragma warning(disable: 4505)
 #  if defined(_DEBUG) && !defined(Py_DEBUG)
+// Workaround for a VS 2022 issue.
+// NOTE: This workaround knowingly violates the Python.h include order requirement:
+// https://docs.python.org/3/c-api/intro.html#include-files
+// See https://github.com/pybind/pybind11/pull/3497 for full context.
+#    include <yvals.h>
+#    if _MSVC_STL_VERSION >= 143
+#      include <crtdefs.h>
+#    endif
 #    define PYBIND11_DEBUG_MARKER
 #    undef _DEBUG
 #  endif
@ -183,6 +191,21 @@
 #  define PYBIND11_HAS_VARIANT 1
 #endif

+#if defined(PYBIND11_CPP17)
+#  if defined(__has_include)
+#    if __has_include(<string_view>)
+#      define PYBIND11_HAS_STRING_VIEW
+#    endif
+#  elif defined(_MSC_VER)
+#    define PYBIND11_HAS_STRING_VIEW
+#  endif
+#endif
+
+#if defined(__cpp_lib_char8_t) && __cpp_lib_char8_t >= 201811L
+#  define PYBIND11_HAS_U8STRING
+#endif
+
+
 #include <Python.h>
 #include <frameobject.h>
 #include <pythread.h>
--- a/include/pybind11/detail/descr.h
+++ b/include/pybind11/detail/descr.h
@ -54,8 +54,8 @@ constexpr descr<N1 + N2, Ts1..., Ts2...> operator+(const descr<N1, Ts1...> &a, c
 }

 template <size_t N>
-constexpr descr<N - 1> _(char const(&text)[N]) { return descr<N - 1>(text); }
-constexpr descr<0> _(char const(&)[1]) { return {}; }
+constexpr descr<N - 1> const_name(char const(&text)[N]) { return descr<N - 1>(text); }
+constexpr descr<0> const_name(char const(&)[1]) { return {}; }

 template <size_t Rem, size_t... Digits> struct int_to_str : int_to_str<Rem/10, Rem%10, Digits...> { };
 template <size_t...Digits> struct int_to_str<0, Digits...> {
@ -64,25 +64,50 @@ template <size_t...Digits> struct int_to_str<0, Digits...> {

 // Ternary description (like std::conditional)
 template <bool B, size_t N1, size_t N2>
-constexpr enable_if_t<B, descr<N1 - 1>> _(char const(&text1)[N1], char const(&)[N2]) {
-    return _(text1);
+constexpr enable_if_t<B, descr<N1 - 1>> const_name(char const(&text1)[N1], char const(&)[N2]) {
+    return const_name(text1);
 }
 template <bool B, size_t N1, size_t N2>
-constexpr enable_if_t<!B, descr<N2 - 1>> _(char const(&)[N1], char const(&text2)[N2]) {
-    return _(text2);
+constexpr enable_if_t<!B, descr<N2 - 1>> const_name(char const(&)[N1], char const(&text2)[N2]) {
+    return const_name(text2);
 }

 template <bool B, typename T1, typename T2>
-constexpr enable_if_t<B, T1> _(const T1 &d, const T2 &) { return d; }
+constexpr enable_if_t<B, T1> const_name(const T1 &d, const T2 &) { return d; }
 template <bool B, typename T1, typename T2>
-constexpr enable_if_t<!B, T2> _(const T1 &, const T2 &d) { return d; }
+constexpr enable_if_t<!B, T2> const_name(const T1 &, const T2 &d) { return d; }

 template <size_t Size>
-auto constexpr _() -> remove_cv_t<decltype(int_to_str<Size / 10, Size % 10>::digits)> {
+auto constexpr const_name() -> remove_cv_t<decltype(int_to_str<Size / 10, Size % 10>::digits)> {
    return int_to_str<Size / 10, Size % 10>::digits;
 }

-template <typename Type> constexpr descr<1, Type> _() { return {'%'}; }
+template <typename Type> constexpr descr<1, Type> const_name() { return {'%'}; }
+
+// The "_" might be defined as a macro - don't define it if so.
+// Repeating the const_name code to avoid introducing a #define.
+#ifndef _
+template <size_t N>
+constexpr descr<N-1> _(char const(&text)[N]) { return const_name<N>(text); }
+template <bool B, size_t N1, size_t N2>
+constexpr enable_if_t<B, descr<N1 - 1>> _(char const(&text1)[N1], char const(&text2)[N2]) {
+    return const_name<B,N1,N2>(text1, text2);
+}
+template <bool B, size_t N1, size_t N2>
+constexpr enable_if_t<!B, descr<N2 - 1>> _(char const(&text1)[N1], char const(&text2)[N2]) {
+    return const_name<B,N1,N2>(text1, text2);
+}
+template <bool B, typename T1, typename T2>
+constexpr enable_if_t<B, T1> _(const T1 &d1, const T2 &d2) { return const_name<B,T1,T2>(d1, d2); }
+template <bool B, typename T1, typename T2>
+constexpr enable_if_t<!B, T2> _(const T1 &d1, const T2 &d2) { return const_name<B,T1,T2>(d1, d2); }
+
+template <size_t Size>
+auto constexpr _() -> remove_cv_t<decltype(int_to_str<Size / 10, Size % 10>::digits)> {
+    return const_name<Size>();
+}
+template <typename Type> constexpr descr<1, Type> _() { return const_name<Type>(); }
+#endif

 constexpr descr<0> concat() { return {}; }

@ -92,12 +117,12 @@ constexpr descr<N, Ts...> concat(const descr<N, Ts...> &descr) { return descr; }
 template <size_t N, typename... Ts, typename... Args>
 constexpr auto concat(const descr<N, Ts...> &d, const Args &...args)
    -> decltype(std::declval<descr<N + 2, Ts...>>() + concat(args...)) {
-    return d + _(", ") + concat(args...);
+    return d + const_name(", ") + concat(args...);
 }

 template <size_t N, typename... Ts>
 constexpr descr<N + 2, Ts...> type_descr(const descr<N, Ts...> &descr) {
-    return _("{") + descr + _("}");
+    return const_name("{") + descr + const_name("}");
 }

 PYBIND11_NAMESPACE_END(detail)
--- a/include/pybind11/detail/init.h
+++ b/include/pybind11/detail/init.h
@ -24,7 +24,7 @@ public:

    template <typename> using cast_op_type = value_and_holder &;
    explicit operator value_and_holder &() { return *value; }
-    static constexpr auto name = _<value_and_holder>();
+    static constexpr auto name = const_name<value_and_holder>();

 private:
    value_and_holder *value = nullptr;
--- a/include/pybind11/detail/type_caster_base.h
+++ b/include/pybind11/detail/type_caster_base.h
@ -897,7 +897,7 @@ template <typename type> class type_caster_base : public type_caster_generic {
    using itype = intrinsic_t<type>;

 public:
-    static constexpr auto name = _<type>();
+    static constexpr auto name = const_name<type>();

    type_caster_base() : type_caster_base(typeid(type)) { }
    explicit type_caster_base(const std::type_info &info) : type_caster_generic(info) { }
--- a/include/pybind11/eigen.h
+++ b/include/pybind11/eigen.h
@ -50,8 +50,12 @@ PYBIND11_NAMESPACE_BEGIN(detail)

 #if EIGEN_VERSION_AT_LEAST(3,3,0)
 using EigenIndex = Eigen::Index;
+template<typename Scalar, int Flags, typename StorageIndex>
+using EigenMapSparseMatrix = Eigen::Map<Eigen::SparseMatrix<Scalar, Flags, StorageIndex>>;
 #else
 using EigenIndex = EIGEN_DEFAULT_DENSE_INDEX_TYPE;
+template<typename Scalar, int Flags, typename StorageIndex>
+using EigenMapSparseMatrix = Eigen::MappedSparseMatrix<Scalar, Flags, StorageIndex>;
 #endif

 // Matches Eigen::Map, Eigen::Ref, blocks, etc:
@ -80,14 +84,12 @@ template <bool EigenRowMajor> struct EigenConformable {
    // Matrix type:
    EigenConformable(EigenIndex r, EigenIndex c,
            EigenIndex rstride, EigenIndex cstride) :
-        conformable{true}, rows{r}, cols{c} {
-        // TODO: when Eigen bug #747 is fixed, remove the tests for non-negativity. http://eigen.tuxfamily.org/bz/show_bug.cgi?id=747
-        if (rstride < 0 || cstride < 0) {
-            negativestrides = true;
-        } else {
-            stride = {EigenRowMajor ? rstride : cstride /* outer stride */,
-                      EigenRowMajor ? cstride : rstride /* inner stride */ };
-        }
+        conformable{true}, rows{r}, cols{c},
+        //TODO: when Eigen bug #747 is fixed, remove the tests for non-negativity. http://eigen.tuxfamily.org/bz/show_bug.cgi?id=747
+        stride{EigenRowMajor ? (rstride > 0 ? rstride : 0) : (cstride > 0 ? cstride : 0) /* outer stride */,
+               EigenRowMajor ? (cstride > 0 ? cstride : 0) : (rstride > 0 ? rstride : 0) /* inner stride */ },
+        negativestrides{rstride < 0 || cstride < 0} {
+
    }
    // Vector type:
    EigenConformable(EigenIndex r, EigenIndex c, EigenIndex stride)
@ -190,20 +192,20 @@ template <typename Type_> struct EigenProps {
    static constexpr bool show_f_contiguous = !show_c_contiguous && show_order && requires_col_major;

    static constexpr auto descriptor =
-        _("numpy.ndarray[") + npy_format_descriptor<Scalar>::name +
-        _("[")  + _<fixed_rows>(_<(size_t) rows>(), _("m")) +
-        _(", ") + _<fixed_cols>(_<(size_t) cols>(), _("n")) +
-        _("]") +
+        const_name("numpy.ndarray[") + npy_format_descriptor<Scalar>::name +
+        const_name("[")  + const_name<fixed_rows>(const_name<(size_t) rows>(), const_name("m")) +
+        const_name(", ") + const_name<fixed_cols>(const_name<(size_t) cols>(), const_name("n")) +
+        const_name("]") +
        // For a reference type (e.g. Ref<MatrixXd>) we have other constraints that might need to be
        // satisfied: writeable=True (for a mutable reference), and, depending on the map's stride
        // options, possibly f_contiguous or c_contiguous.  We include them in the descriptor output
        // to provide some hint as to why a TypeError is occurring (otherwise it can be confusing to
        // see that a function accepts a 'numpy.ndarray[float64[3,2]]' and an error message that you
        // *gave* a numpy.ndarray of the right type and dimensions.
-        _<show_writeable>(", flags.writeable", "") +
-        _<show_c_contiguous>(", flags.c_contiguous", "") +
-        _<show_f_contiguous>(", flags.f_contiguous", "") +
-        _("]");
+        const_name<show_writeable>(", flags.writeable", "") +
+        const_name<show_c_contiguous>(", flags.c_contiguous", "") +
+        const_name<show_f_contiguous>(", flags.f_contiguous", "") +
+        const_name("]");
 };

 // Casts an Eigen type to numpy array.  If given a base, the numpy array references the src data,
@ -573,7 +575,7 @@ struct type_caster<Type, enable_if_t<is_eigen_sparse<Type>::value>> {
        if (!values || !innerIndices || !outerIndices)
            return false;

-        value = Eigen::MappedSparseMatrix<Scalar,
+        value = EigenMapSparseMatrix<Scalar,
                                     Type::Flags & (Eigen::RowMajor | Eigen::ColMajor),
                                     StorageIndex>(
            shape[0].cast<Index>(), shape[1].cast<Index>(), nnz,
@ -598,8 +600,8 @@ struct type_caster<Type, enable_if_t<is_eigen_sparse<Type>::value>> {
        ).release();
    }

-    PYBIND11_TYPE_CASTER(Type, _<(Type::IsRowMajor) != 0>("scipy.sparse.csr_matrix[", "scipy.sparse.csc_matrix[")
-            + npy_format_descriptor<Scalar>::name + _("]"));
+    PYBIND11_TYPE_CASTER(Type, const_name<(Type::IsRowMajor) != 0>("scipy.sparse.csr_matrix[", "scipy.sparse.csc_matrix[")
+            + npy_format_descriptor<Scalar>::name + const_name("]"));
 };

 PYBIND11_NAMESPACE_END(detail)
--- a/include/pybind11/embed.h
+++ b/include/pybind11/embed.h
@ -102,6 +102,13 @@ inline wchar_t *widen_chars(const char *safe_arg) {
    wchar_t *widened_arg = Py_DecodeLocale(safe_arg, nullptr);
 #else
    wchar_t *widened_arg = nullptr;
+
+// warning C4996: 'mbstowcs': This function or variable may be unsafe.
+#if defined(_MSC_VER)
+#pragma warning(push)
+#pragma warning(disable:4996)
+#endif
+
 #    if defined(HAVE_BROKEN_MBSTOWCS) && HAVE_BROKEN_MBSTOWCS
    size_t count = strlen(safe_arg);
 #    else
@ -111,6 +118,11 @@ inline wchar_t *widen_chars(const char *safe_arg) {
        widened_arg = new wchar_t[count + 1];
        mbstowcs(widened_arg, safe_arg, count + 1);
    }
+
+#if defined(_MSC_VER)
+#pragma warning(pop)
+#endif
+
 #endif
    return widened_arg;
 }
--- a/include/pybind11/eval.h
+++ b/include/pybind11/eval.h
@ -19,11 +19,11 @@ PYBIND11_NAMESPACE_BEGIN(PYBIND11_NAMESPACE)
 PYBIND11_NAMESPACE_BEGIN(detail)

 inline void ensure_builtins_in_globals(object &global) {
-    #if PY_VERSION_HEX < 0x03080000
+    #if defined(PYPY_VERSION) || PY_VERSION_HEX < 0x03080000
        // Running exec and eval on Python 2 and 3 adds `builtins` module under
        // `__builtins__` key to globals if not yet present.
        // Python 3.8 made PyRun_String behave similarly. Let's also do that for
-        // older versions, for consistency.
+        // older versions, for consistency. This was missing from PyPy3.8 7.3.7.
        if (!global.contains("__builtins__"))
            global["__builtins__"] = module_::import(PYBIND11_BUILTINS_MODULE);
    #else
--- a/include/pybind11/functional.h
+++ b/include/pybind11/functional.h
@ -113,8 +113,8 @@ public:
        return cpp_function(std::forward<Func>(f_), policy).release();
    }

-    PYBIND11_TYPE_CASTER(type, _("Callable[[") + concat(make_caster<Args>::name...) + _("], ")
-                               + make_caster<retval_type>::name + _("]"));
+    PYBIND11_TYPE_CASTER(type, const_name("Callable[[") + concat(make_caster<Args>::name...) + const_name("], ")
+                               + make_caster<retval_type>::name + const_name("]"));
 };

 PYBIND11_NAMESPACE_END(detail)
--- a/include/pybind11/numpy.h
+++ b/include/pybind11/numpy.h
@ -39,7 +39,7 @@ class array; // Forward declaration

 PYBIND11_NAMESPACE_BEGIN(detail)

-template <> struct handle_type_name<array> { static constexpr auto name = _("numpy.ndarray"); };
+template <> struct handle_type_name<array> { static constexpr auto name = const_name("numpy.ndarray"); };

 template <typename type, typename SFINAE = void> struct npy_format_descriptor;

@ -290,7 +290,7 @@ template <typename T> struct array_info_scalar {
    using type = T;
    static constexpr bool is_array = false;
    static constexpr bool is_empty = false;
-    static constexpr auto extents = _("");
+    static constexpr auto extents = const_name("");
    static void append_extents(list& /* shape */) { }
 };
 // Computes underlying type and a comma-separated list of extents for array
@ -309,8 +309,8 @@ template <typename T, size_t N> struct array_info<std::array<T, N>> {
        array_info<T>::append_extents(shape);
    }

-    static constexpr auto extents = _<array_info<T>::is_array>(
-        concat(_<N>(), array_info<T>::extents), _<N>()
+    static constexpr auto extents = const_name<array_info<T>::is_array>(
+        concat(const_name<N>(), array_info<T>::extents), const_name<N>()
    );
 };
 // For numpy we have special handling for arrays of characters, so we don't include
@ -1021,7 +1021,7 @@ template <typename T>
 struct format_descriptor<T, detail::enable_if_t<detail::array_info<T>::is_array>> {
    static std::string format() {
        using namespace detail;
-        static constexpr auto extents = _("(") + array_info<T>::extents + _(")");
+        static constexpr auto extents = const_name("(") + array_info<T>::extents + const_name(")");
        return extents.text + format_descriptor<remove_all_extents_t<T>>::format();
    }
 };
@ -1056,28 +1056,28 @@ struct npy_format_descriptor_name;

 template <typename T>
 struct npy_format_descriptor_name<T, enable_if_t<std::is_integral<T>::value>> {
-    static constexpr auto name = _<std::is_same<T, bool>::value>(
-        _("bool"), _<std::is_signed<T>::value>("numpy.int", "numpy.uint") + _<sizeof(T)*8>()
+    static constexpr auto name = const_name<std::is_same<T, bool>::value>(
+        const_name("bool"), const_name<std::is_signed<T>::value>("numpy.int", "numpy.uint") + const_name<sizeof(T)*8>()
    );
 };

 template <typename T>
 struct npy_format_descriptor_name<T, enable_if_t<std::is_floating_point<T>::value>> {
-    static constexpr auto name = _<std::is_same<T, float>::value
+    static constexpr auto name = const_name<std::is_same<T, float>::value
                                   || std::is_same<T, const float>::value
                                   || std::is_same<T, double>::value
                                   || std::is_same<T, const double>::value>(
-        _("numpy.float") + _<sizeof(T)*8>(), _("numpy.longdouble")
+        const_name("numpy.float") + const_name<sizeof(T)*8>(), const_name("numpy.longdouble")
    );
 };

 template <typename T>
 struct npy_format_descriptor_name<T, enable_if_t<is_complex<T>::value>> {
-    static constexpr auto name = _<std::is_same<typename T::value_type, float>::value
+    static constexpr auto name = const_name<std::is_same<typename T::value_type, float>::value
                                   || std::is_same<typename T::value_type, const float>::value
                                   || std::is_same<typename T::value_type, double>::value
                                   || std::is_same<typename T::value_type, const double>::value>(
-        _("numpy.complex") + _<sizeof(typename T::value_type)*16>(), _("numpy.longcomplex")
+        const_name("numpy.complex") + const_name<sizeof(typename T::value_type)*16>(), const_name("numpy.longcomplex")
    );
 };

@ -1105,7 +1105,7 @@ public:
 };

 #define PYBIND11_DECL_CHAR_FMT \
-    static constexpr auto name = _("S") + _<N>(); \
+    static constexpr auto name = const_name("S") + const_name<N>(); \
    static pybind11::dtype dtype() { return pybind11::dtype(std::string("S") + std::to_string(N)); }
 template <size_t N> struct npy_format_descriptor<char[N]> { PYBIND11_DECL_CHAR_FMT };
 template <size_t N> struct npy_format_descriptor<std::array<char, N>> { PYBIND11_DECL_CHAR_FMT };
@ -1117,7 +1117,7 @@ private:
 public:
    static_assert(!array_info<T>::is_empty, "Zero-sized arrays are not supported");

-    static constexpr auto name = _("(") + array_info<T>::extents + _(")") + base_descr::name;
+    static constexpr auto name = const_name("(") + array_info<T>::extents + const_name(")") + base_descr::name;
    static pybind11::dtype dtype() {
        list shape;
        array_info<T>::append_extents(shape);
@ -1705,7 +1705,7 @@ vectorize_extractor(const Func &f, Return (*) (Args ...)) {
 }

 template <typename T, int Flags> struct handle_type_name<array_t<T, Flags>> {
-    static constexpr auto name = _("numpy.ndarray[") + npy_format_descriptor<T>::name + _("]");
+    static constexpr auto name = const_name("numpy.ndarray[") + npy_format_descriptor<T>::name + const_name("]");
 };

 PYBIND11_NAMESPACE_END(detail)
--- a/include/pybind11/pybind11.h
+++ b/include/pybind11/pybind11.h
@ -115,7 +115,7 @@ public:
    template <typename Return, typename Class, typename... Arg, typename... Extra>
    // NOLINTNEXTLINE(google-explicit-constructor)
    cpp_function(Return (Class::*f)(Arg...)&, const Extra&... extra) {
-        initialize([f](Class *c, Arg... args) -> Return { return (c->*f)(args...); },
+        initialize([f](Class *c, Arg... args) -> Return { return (c->*f)(std::forward<Arg>(args)...); },
                   (Return (*) (Class *, Arg...)) nullptr, extra...);
    }

@ -133,7 +133,7 @@ public:
    template <typename Return, typename Class, typename... Arg, typename... Extra>
    // NOLINTNEXTLINE(google-explicit-constructor)
    cpp_function(Return (Class::*f)(Arg...) const&, const Extra&... extra) {
-        initialize([f](const Class *c, Arg... args) -> Return { return (c->*f)(args...); },
+        initialize([f](const Class *c, Arg... args) -> Return { return (c->*f)(std::forward<Arg>(args)...); },
                   (Return (*)(const Class *, Arg ...)) nullptr, extra...);
    }

@ -203,7 +203,7 @@ protected:
            conditional_t<std::is_void<Return>::value, void_type, Return>
        >;

-        static_assert(expected_num_args<Extra...>(sizeof...(Args), cast_in::has_args, cast_in::has_kwargs),
+        static_assert(expected_num_args<Extra...>(sizeof...(Args), cast_in::args_pos >= 0, cast_in::has_kwargs),
                      "The number of argument annotations does not match the number of function arguments");

        /* Dispatch code which converts function arguments and performs the actual function call */
@ -238,30 +238,37 @@ protected:
            return result;
        };

+        rec->nargs_pos = cast_in::args_pos >= 0
+            ? static_cast<std::uint16_t>(cast_in::args_pos)
+            : sizeof...(Args) - cast_in::has_kwargs; // Will get reduced more if we have a kw_only
+        rec->has_args = cast_in::args_pos >= 0;
+        rec->has_kwargs = cast_in::has_kwargs;
+
        /* Process any user-provided function attributes */
        process_attributes<Extra...>::init(extra..., rec);

        {
            constexpr bool has_kw_only_args = any_of<std::is_same<kw_only, Extra>...>::value,
                           has_pos_only_args = any_of<std::is_same<pos_only, Extra>...>::value,
-                           has_args = any_of<std::is_same<args, Args>...>::value,
                           has_arg_annotations = any_of<is_keyword<Extra>...>::value;
            static_assert(has_arg_annotations || !has_kw_only_args, "py::kw_only requires the use of argument annotations");
            static_assert(has_arg_annotations || !has_pos_only_args, "py::pos_only requires the use of argument annotations (for docstrings and aligning the annotations to the argument)");
-            static_assert(!(has_args && has_kw_only_args), "py::kw_only cannot be combined with a py::args argument");
+
+            static_assert(constexpr_sum(is_kw_only<Extra>::value...) <= 1, "py::kw_only may be specified only once");
+            static_assert(constexpr_sum(is_pos_only<Extra>::value...) <= 1, "py::pos_only may be specified only once");
+            constexpr auto kw_only_pos = constexpr_first<is_kw_only, Extra...>();
+            constexpr auto pos_only_pos = constexpr_first<is_pos_only, Extra...>();
+            static_assert(!(has_kw_only_args && has_pos_only_args) || pos_only_pos < kw_only_pos, "py::pos_only must come before py::kw_only");
        }

        /* Generate a readable signature describing the function's arguments and return value types */
-        static constexpr auto signature = _("(") + cast_in::arg_names + _(") -> ") + cast_out::name;
+        static constexpr auto signature = const_name("(") + cast_in::arg_names + const_name(") -> ") + cast_out::name;
        PYBIND11_DESCR_CONSTEXPR auto types = decltype(signature)::types();

        /* Register the function with Python from generic (non-templated) code */
        // Pass on the ownership over the `unique_rec` to `initialize_generic`. `rec` stays valid.
        initialize_generic(std::move(unique_rec), signature.text, types.data(), sizeof...(Args));

-        if (cast_in::has_args) rec->has_args = true;
-        if (cast_in::has_kwargs) rec->has_kwargs = true;
-
        /* Stash some additional information used by an important optimization in 'functional.h' */
        using FunctionType = Return (*)(Args...);
        constexpr bool is_function_ptr =
@ -340,16 +347,18 @@ protected:
        /* Generate a proper function signature */
        std::string signature;
        size_t type_index = 0, arg_index = 0;
+        bool is_starred = false;
        for (auto *pc = text; *pc != '\0'; ++pc) {
            const auto c = *pc;

            if (c == '{') {
                // Write arg name for everything except *args and **kwargs.
-                if (*(pc + 1) == '*')
+                is_starred = *(pc + 1) == '*';
+                if (is_starred)
                    continue;
                // Separator for keyword-only arguments, placed before the kw
-                // arguments start
-                if (rec->nargs_kw_only > 0 && arg_index + rec->nargs_kw_only == args)
+                // arguments start (unless we are already putting an *args)
+                if (!rec->has_args && arg_index == rec->nargs_pos)
                    signature += "*, ";
                if (arg_index < rec->args.size() && rec->args[arg_index].name) {
                    signature += rec->args[arg_index].name;
@ -361,7 +370,7 @@ protected:
                signature += ": ";
            } else if (c == '}') {
                // Write default value if available.
-                if (arg_index < rec->args.size() && rec->args[arg_index].descr) {
+                if (!is_starred && arg_index < rec->args.size() && rec->args[arg_index].descr) {
                    signature += " = ";
                    signature += rec->args[arg_index].descr;
                }
@ -369,6 +378,7 @@ protected:
                // argument, rather than before like *
                if (rec->nargs_pos_only > 0 && (arg_index + 1) == rec->nargs_pos_only)
                    signature += ", /";
+                if (!is_starred)
                    arg_index++;
            } else if (c == '%') {
                const std::type_info *t = types[type_index++];
@ -395,7 +405,7 @@ protected:
            }
        }

-        if (arg_index != args || types[type_index] != nullptr)
+        if (arg_index != args - rec->has_args - rec->has_kwargs || types[type_index] != nullptr)
            pybind11_fail("Internal error while parsing type signature (2)");

 #if PY_MAJOR_VERSION < 3
@ -631,7 +641,7 @@ protected:
                      named positional arguments weren't *also* specified via kwarg.
                   2. If we weren't given enough, try to make up the omitted ones by checking
                      whether they were provided by a kwarg matching the `py::arg("name")` name.  If
-                      so, use it (and remove it from kwargs; if not, see if the function binding
+                      so, use it (and remove it from kwargs); if not, see if the function binding
                      provided a default that we can use.
                   3. Ensure that either all keyword arguments were "consumed", or that the function
                      takes a kwargs argument to accept unconsumed kwargs.
@ -649,7 +659,7 @@ protected:
                size_t num_args = func.nargs;    // Number of positional arguments that we need
                if (func.has_args) --num_args;   // (but don't count py::args
                if (func.has_kwargs) --num_args; //  or py::kwargs)
-                size_t pos_args = num_args - func.nargs_kw_only;
+                size_t pos_args = func.nargs_pos;

                if (!func.has_args && n_args_in > pos_args)
                    continue; // Too many positional arguments for this overload
@ -695,6 +705,10 @@ protected:
                if (bad_arg)
                    continue; // Maybe it was meant for another overload (issue #688)

+                // Keep track of how many position args we copied out in case we need to come back
+                // to copy the rest into a py::args argument.
+                size_t positional_args_copied = args_copied;
+
                // We'll need to copy this if we steal some kwargs for defaults
                dict kwargs = reinterpret_borrow<dict>(kwargs_in);

@ -747,6 +761,10 @@ protected:
                        }

                        if (value) {
+                            // If we're at the py::args index then first insert a stub for it to be replaced later
+                            if (func.has_args && call.args.size() == func.nargs_pos)
+                                call.args.push_back(none());
+
                            call.args.push_back(value);
                            call.args_convert.push_back(arg_rec.convert);
                        }
@ -769,16 +787,19 @@ protected:
                        // We didn't copy out any position arguments from the args_in tuple, so we
                        // can reuse it directly without copying:
                        extra_args = reinterpret_borrow<tuple>(args_in);
-                    } else if (args_copied >= n_args_in) {
+                    } else if (positional_args_copied >= n_args_in) {
                        extra_args = tuple(0);
                    } else {
-                        size_t args_size = n_args_in - args_copied;
+                        size_t args_size = n_args_in - positional_args_copied;
                        extra_args = tuple(args_size);
                        for (size_t i = 0; i < args_size; ++i) {
-                            extra_args[i] = PyTuple_GET_ITEM(args_in, args_copied + i);
+                            extra_args[i] = PyTuple_GET_ITEM(args_in, positional_args_copied + i);
                        }
                    }
+                    if (call.args.size() <= func.nargs_pos)
                        call.args.push_back(extra_args);
+                    else
+                        call.args[func.nargs_pos] = extra_args;
                    call.args_convert.push_back(false);
                    call.args_ref = std::move(extra_args);
                }
@ -1958,6 +1979,16 @@ inline std::pair<decltype(internals::registered_types_py)::iterator, bool> all_t
        // gets destroyed:
        weakref((PyObject *) type, cpp_function([type](handle wr) {
            get_internals().registered_types_py.erase(type);
+
+            // TODO consolidate the erasure code in pybind11_meta_dealloc() in class.h
+            auto &cache = get_internals().inactive_override_cache;
+            for (auto it = cache.begin(), last = cache.end(); it != last; ) {
+                if (it->first == reinterpret_cast<PyObject *>(type))
+                    it = cache.erase(it);
+                else
+                    ++it;
+            }
+
            wr.dec_ref();
        })).release();
    }
@ -2334,8 +2365,9 @@ inline function get_type_override(const void *this_ptr, const type_info *this_ty

    /* Don't call dispatch code if invoked from overridden function.
       Unfortunately this doesn't work on PyPy. */
-#if !defined(PYPY_VERSION)
-
+#if !defined(PYPY_VERSION) && PY_VERSION_HEX < 0x030B0000
+    // TODO: Remove PyPy workaround for Python 3.11.
+    // Current API fails on 3.11 since co_varnames can be null.
 #if PY_VERSION_HEX >= 0x03090000
    PyFrameObject *frame = PyThreadState_GetFrame(PyThreadState_Get());
    if (frame != nullptr) {
@ -2343,7 +2375,7 @@ inline function get_type_override(const void *this_ptr, const type_info *this_ty
        // f_code is guaranteed to not be NULL
        if ((std::string) str(f_code->co_name) == name && f_code->co_argcount > 0) {
            PyObject* locals = PyEval_GetLocals();
-            if (locals != nullptr) {
+            if (locals != nullptr && f_code->co_varnames != nullptr) {
                PyObject *self_caller = dict_getitem(
                    locals, PyTuple_GET_ITEM(f_code->co_varnames, 0)
                );
--- a/include/pybind11/pytypes.h
+++ b/include/pybind11/pytypes.h
@ -18,6 +18,10 @@
 #  include <optional>
 #endif

+#ifdef PYBIND11_HAS_STRING_VIEW
+#  include <string_view>
+#endif
+
 PYBIND11_NAMESPACE_BEGIN(PYBIND11_NAMESPACE)

 /* A few forward declarations */
@ -287,10 +291,10 @@ protected:
    struct borrowed_t { };
    struct stolen_t { };

-#ifndef DOXYGEN_SHOULD_SKIP_THIS  // Issue in breathe 4.26.1
+    /// @cond BROKEN
    template <typename T> friend T reinterpret_borrow(handle);
    template <typename T> friend T reinterpret_steal(handle);
-#endif
+    /// @endcond

 public:
    // Only accessible from derived classes and the reinterpret_* functions
@ -431,7 +435,7 @@ inline void raise_from(error_already_set& err, PyObject *type, const char *messa

 #endif

-/** \defgroup python_builtins _
+/** \defgroup python_builtins const_name
    Unless stated otherwise, the following C++ functions behave the same
    as their Python counterparts.
 */
@ -1085,6 +1089,20 @@ public:
    // NOLINTNEXTLINE(google-explicit-constructor)
    str(const std::string &s) : str(s.data(), s.size()) { }

+#ifdef PYBIND11_HAS_STRING_VIEW
+    // enable_if is needed to avoid "ambiguous conversion" errors (see PR #3521).
+    template <typename T, detail::enable_if_t<std::is_same<T, std::string_view>::value, int> = 0>
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    str(T s) : str(s.data(), s.size()) { }
+
+# ifdef PYBIND11_HAS_U8STRING
+    // reinterpret_cast here is safe (C++20 guarantees char8_t has the same size/alignment as char)
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    str(std::u8string_view s) : str(reinterpret_cast<const char*>(s.data()), s.size()) { }
+# endif
+
+#endif
+
    explicit str(const bytes &b);

    /** \rst
@ -1167,6 +1185,26 @@ public:
            pybind11_fail("Unable to extract bytes contents!");
        return std::string(buffer, (size_t) length);
    }
+
+#ifdef PYBIND11_HAS_STRING_VIEW
+    // enable_if is needed to avoid "ambiguous conversion" errors (see PR #3521).
+    template <typename T, detail::enable_if_t<std::is_same<T, std::string_view>::value, int> = 0>
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    bytes(T s) : bytes(s.data(), s.size()) { }
+
+    // Obtain a string view that views the current `bytes` buffer value.  Note that this is only
+    // valid so long as the `bytes` instance remains alive and so generally should not outlive the
+    // lifetime of the `bytes` instance.
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    operator std::string_view() const {
+        char *buffer = nullptr;
+        ssize_t length = 0;
+        if (PYBIND11_BYTES_AS_STRING_AND_SIZE(m_ptr, &buffer, &length))
+            pybind11_fail("Unable to extract bytes contents!");
+        return {buffer, static_cast<size_t>(length)};
+    }
+#endif
+
 };
 // Note: breathe >= 4.17.0 will fail to build docs if the below two constructors
 // are included in the doxygen group; close here and reopen after as a workaround
@ -1714,10 +1752,17 @@ public:
    static memoryview from_memory(const void *mem, ssize_t size) {
        return memoryview::from_memory(const_cast<void*>(mem), size, true);
    }
+
+#ifdef PYBIND11_HAS_STRING_VIEW
+    static memoryview from_memory(std::string_view mem) {
+        return from_memory(const_cast<char*>(mem.data()), static_cast<ssize_t>(mem.size()), true);
+    }
+#endif
+
 #endif
 };

-#ifndef DOXYGEN_SHOULD_SKIP_THIS
+/// @cond DUPLICATE
 inline memoryview memoryview::from_buffer(
    void *ptr, ssize_t itemsize, const char* format,
    detail::any_container<ssize_t> shape,
@ -1745,7 +1790,7 @@ inline memoryview memoryview::from_buffer(
        throw error_already_set();
    return memoryview(object(obj, stolen_t{}));
 }
-#endif  // DOXYGEN_SHOULD_SKIP_THIS
+/// @endcond
 /// @} pytypes

 /// \addtogroup python_builtins
--- a/include/pybind11/stl.h
+++ b/include/pybind11/stl.h
@ -78,7 +78,7 @@ template <typename Type, typename Key> struct set_caster {
        return s.release();
    }

-    PYBIND11_TYPE_CASTER(type, _("Set[") + key_conv::name + _("]"));
+    PYBIND11_TYPE_CASTER(type, const_name("Set[") + key_conv::name + const_name("]"));
 };

 template <typename Type, typename Key, typename Value> struct map_caster {
@ -120,7 +120,7 @@ template <typename Type, typename Key, typename Value> struct map_caster {
        return d.release();
    }

-    PYBIND11_TYPE_CASTER(Type, _("Dict[") + key_conv::name + _(", ") + value_conv::name + _("]"));
+    PYBIND11_TYPE_CASTER(Type, const_name("Dict[") + key_conv::name + const_name(", ") + value_conv::name + const_name("]"));
 };

 template <typename Type, typename Value> struct list_caster {
@ -166,7 +166,7 @@ public:
        return l.release();
    }

-    PYBIND11_TYPE_CASTER(Type, _("List[") + value_conv::name + _("]"));
+    PYBIND11_TYPE_CASTER(Type, const_name("List[") + value_conv::name + const_name("]"));
 };

 template <typename Type, typename Alloc> struct type_caster<std::vector<Type, Alloc>>
@ -223,7 +223,7 @@ public:
        return l.release();
    }

-    PYBIND11_TYPE_CASTER(ArrayType, _("List[") + value_conv::name + _<Resizable>(_(""), _("[") + _<Size>() + _("]")) + _("]"));
+    PYBIND11_TYPE_CASTER(ArrayType, const_name("List[") + value_conv::name + const_name<Resizable>(const_name(""), const_name("[") + const_name<Size>() + const_name("]")) + const_name("]"));
 };

 template <typename Type, size_t Size> struct type_caster<std::array<Type, Size>>
@ -273,7 +273,7 @@ template<typename Type, typename Value = typename Type::value_type> struct optio
        return true;
    }

-    PYBIND11_TYPE_CASTER(Type, _("Optional[") + value_conv::name + _("]"));
+    PYBIND11_TYPE_CASTER(Type, const_name("Optional[") + value_conv::name + const_name("]"));
 };

 #if defined(PYBIND11_HAS_OPTIONAL)
@ -353,7 +353,7 @@ struct variant_caster<V<Ts...>> {
    }

    using Type = V<Ts...>;
-    PYBIND11_TYPE_CASTER(Type, _("Union[") + detail::concat(make_caster<Ts>::name...) + _("]"));
+    PYBIND11_TYPE_CASTER(Type, const_name("Union[") + detail::concat(make_caster<Ts>::name...) + const_name("]"));
 };

 #if defined(PYBIND11_HAS_VARIANT)
--- a/include/pybind11/stl/filesystem.h
+++ b/include/pybind11/stl/filesystem.h
@ -92,7 +92,7 @@ public:
        return true;
    }

-    PYBIND11_TYPE_CASTER(T, _("os.PathLike"));
+    PYBIND11_TYPE_CASTER(T, const_name("os.PathLike"));
 };

 template<> struct type_caster<std::filesystem::path>
--- a/noxfile.py
+++ b/noxfile.py
@ -57,10 +57,10 @@ def docs(session: nox.Session) -> None:
    session.chdir("docs")

    if "pdf" in session.posargs:
-        session.run("sphinx-build", "-M", "latexpdf", ".", "_build")
+        session.run("sphinx-build", "-b", "latexpdf", ".", "_build")
        return

-    session.run("sphinx-build", "-M", "html", ".", "_build")
+    session.run("sphinx-build", "-b", "html", ".", "_build")

    if "serve" in session.posargs:
        session.log("Launching docs at http://localhost:8000/ - use Ctrl-C to quit")
@ -85,5 +85,9 @@ def build(session: nox.Session) -> None:
    """

    session.install("build")
-    session.run("python", "-m", "build")
-    session.run("python", "-m", "build", env={"PYBIND11_GLOBAL_SDIST": "1"})
+    session.log("Building normal files")
+    session.run("python", "-m", "build", *session.posargs)
+    session.log("Building pybind11-global files (PYBIND11_GLOBAL_SDIST=1)")
+    session.run(
+        "python", "-m", "build", *session.posargs, env={"PYBIND11_GLOBAL_SDIST": "1"}
+    )
--- a/pybind11/_version.py
+++ b/pybind11/_version.py
@ -8,5 +8,5 @@ def _to_int(s):
        return s


-__version__ = "2.8.1"
+__version__ = "2.9.0"
 version_info = tuple(_to_int(s) for s in __version__.split("."))
--- a/pybind11/setup_helpers.py
+++ b/pybind11/setup_helpers.py
@ -42,6 +42,7 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 import contextlib
 import os
 import platform
+import shlex
 import shutil
 import sys
 import sysconfig
@ -143,7 +144,12 @@ class Pybind11Extension(_Extension):
        if WIN:
            cflags += ["/EHsc", "/bigobj"]
        else:
-            cflags += ["-fvisibility=hidden", "-g0"]
+            cflags += ["-fvisibility=hidden"]
+            env_cflags = os.environ.get("CFLAGS", "")
+            env_cppflags = os.environ.get("CPPFLAGS", "")
+            c_cpp_flags = shlex.split(env_cflags) + shlex.split(env_cppflags)
+            if not any(opt.startswith("-g") for opt in c_cpp_flags):
+                cflags += ["-g0"]
            if MACOS:
                cflags += ["-stdlib=libc++"]
                ldflags += ["-stdlib=libc++"]
@ -460,8 +466,14 @@ class ParallelCompile(object):
                    threads = 1

            if threads > 1:
-                for _ in ThreadPool(threads).imap_unordered(_single_compile, objects):
+                pool = ThreadPool(threads)
+                # In Python 2, ThreadPool can't be used as a context manager.
+                # Once we are no longer supporting it, this can be 'with pool:'
+                try:
+                    for _ in pool.imap_unordered(_single_compile, objects):
                        pass
+                finally:
+                    pool.terminate()
            else:
                for ob in objects:
                    _single_compile(ob)
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@ -429,6 +429,14 @@ foreach(target ${test_targets})
  endif()
 endforeach()

+# Provide nice organisation in IDEs
+if(NOT CMAKE_VERSION VERSION_LESS 3.8)
+  source_group(
+    TREE "${CMAKE_CURRENT_SOURCE_DIR}/../include"
+    PREFIX "Header Files"
+    FILES ${PYBIND11_HEADERS})
+endif()
+
 # Make sure pytest is found or produce a warning
 pybind11_find_import(pytest VERSION 3.1)

--- a/tests/extra_setuptools/test_setuphelper.py
+++ b/tests/extra_setuptools/test_setuphelper.py
@ -8,6 +8,7 @@ import pytest

 DIR = os.path.abspath(os.path.dirname(__file__))
 MAIN_DIR = os.path.dirname(os.path.dirname(DIR))
+WIN = sys.platform.startswith("win32") or sys.platform.startswith("cygwin")


@pytest.mark.parametrize("parallel", [False, True])
@ -71,13 +72,20 @@ def test_simple_setup_py(monkeypatch, tmpdir, parallel, std):
        encoding="ascii",
    )

-    subprocess.check_call(
+    out = subprocess.check_output(
        [sys.executable, "setup.py", "build_ext", "--inplace"],
-        stdout=sys.stdout,
-        stderr=sys.stderr,
    )
+    if not WIN:
+        assert b"-g0" in out
+    out = subprocess.check_output(
+        [sys.executable, "setup.py", "build_ext", "--inplace", "--force"],
+        env=dict(os.environ, CFLAGS="-g"),
+    )
+    if not WIN:
+        assert b"-g0" not in out

    # Debug helper printout, normally hidden
+    print(out)
    for item in tmpdir.listdir():
        print(item.basename)

--- a/tests/pybind11_tests.h
+++ b/tests/pybind11_tests.h
@ -65,7 +65,7 @@ PYBIND11_NAMESPACE_BEGIN(pybind11)
 PYBIND11_NAMESPACE_BEGIN(detail)
 template<> class type_caster<RValueCaster> {
 public:
-    PYBIND11_TYPE_CASTER(RValueCaster, _("RValueCaster"));
+    PYBIND11_TYPE_CASTER(RValueCaster, const_name("RValueCaster"));
    static handle cast(RValueCaster &&, return_value_policy, handle) { return py::str("rvalue").release(); }
    static handle cast(const RValueCaster &, return_value_policy, handle) { return py::str("lvalue").release(); }
 };
--- a/tests/requirements.txt
+++ b/tests/requirements.txt
@ -2,7 +2,7 @@ numpy==1.16.6; python_version<"3.6" and sys_platform!="win32" and platform_pytho
 numpy==1.19.0; platform_python_implementation=="PyPy" and sys_platform=="linux" and python_version=="3.6"
 numpy==1.20.0; platform_python_implementation=="PyPy" and sys_platform=="linux" and python_version=="3.7"
 numpy==1.19.3; platform_python_implementation!="PyPy" and python_version=="3.6"
-numpy==1.21.3; platform_python_implementation!="PyPy" and python_version>="3.7"
+numpy==1.21.3; platform_python_implementation!="PyPy" and python_version>="3.7" and python_version<"3.11"
 py @ git+https://github.com/pytest-dev/py; python_version>="3.11"
 pytest==4.6.9; python_version<"3.5"
 pytest==6.1.2; python_version=="3.5"
--- a/tests/test_buffers.py
+++ b/tests/test_buffers.py
@ -36,6 +36,10 @@ def test_from_python():


 # https://foss.heptapod.net/pypy/pypy/-/issues/2444
+# TODO: fix on recent PyPy
+@pytest.mark.xfail(
+    env.PYPY, reason="PyPy 7.3.7 doesn't clear this anymore", strict=False
+)
 def test_to_python():
    mat = m.Matrix(5, 4)
    assert memoryview(mat).shape == (5, 4)
--- a/tests/test_builtin_casters.cpp
+++ b/tests/test_builtin_casters.cpp
@ -19,7 +19,7 @@ PYBIND11_NAMESPACE_BEGIN(detail)
 template <>
 class type_caster<ConstRefCasted> {
 public:
-  static constexpr auto name = _<ConstRefCasted>();
+  static constexpr auto name = const_name<ConstRefCasted>();

  // Input is unimportant, a new value will always be constructed based on the
  // cast operator.
@ -140,11 +140,35 @@ TEST_SUBMODULE(builtin_casters, m) {
    m.def("string_view16_return", []() { return std::u16string_view(u"utf16 secret \U0001f382"); });
    m.def("string_view32_return", []() { return std::u32string_view(U"utf32 secret \U0001f382"); });

+    // The inner lambdas here are to also test implicit conversion
+    using namespace std::literals;
+    m.def("string_view_bytes", []() { return [](py::bytes b) { return b; }("abc \x80\x80 def"sv); });
+    m.def("string_view_str",   []() { return [](py::str s) { return s; }("abc \342\200\275 def"sv); });
+    m.def("string_view_from_bytes", [](const py::bytes &b) { return [](std::string_view s) { return s; }(b); });
+#if PY_MAJOR_VERSION >= 3
+    m.def("string_view_memoryview", []() {
+        static constexpr auto val = "Have some \360\237\216\202"sv;
+        return py::memoryview::from_memory(val);
+    });
+#endif
+
 #   ifdef PYBIND11_HAS_U8STRING
    m.def("string_view8_print",  [](std::u8string_view s) { py::print(s, s.size()); });
    m.def("string_view8_chars",  [](std::u8string_view s) { py::list l; for (auto c : s) l.append((std::uint8_t) c); return l; });
    m.def("string_view8_return", []() { return std::u8string_view(u8"utf8 secret \U0001f382"); });
+    m.def("string_view8_str",    []() { return py::str{std::u8string_view{u8"abc ‽ def"}}; });
 #   endif
+
+    struct TypeWithBothOperatorStringAndStringView {
+        // NOLINTNEXTLINE(google-explicit-constructor)
+        operator std::string() const { return "success"; }
+        // NOLINTNEXTLINE(google-explicit-constructor)
+        operator std::string_view() const { return "failure"; }
+    };
+    m.def("bytes_from_type_with_both_operator_string_and_string_view",
+          []() { return py::bytes(TypeWithBothOperatorStringAndStringView()); });
+    m.def("str_from_type_with_both_operator_string_and_string_view",
+          []() { return py::str(TypeWithBothOperatorStringAndStringView()); });
 #endif

    // test_integer_casting
--- a/tests/test_builtin_casters.py
+++ b/tests/test_builtin_casters.py
@ -206,6 +206,17 @@ def test_string_view(capture):
        """
        )

+    assert m.string_view_bytes() == b"abc \x80\x80 def"
+    assert m.string_view_str() == u"abc ‽ def"
+    assert m.string_view_from_bytes(u"abc ‽ def".encode("utf-8")) == u"abc ‽ def"
+    if hasattr(m, "has_u8string"):
+        assert m.string_view8_str() == u"abc ‽ def"
+    if not env.PY2:
+        assert m.string_view_memoryview() == "Have some 🎂".encode()
+
+    assert m.bytes_from_type_with_both_operator_string_and_string_view() == b"success"
+    assert m.str_from_type_with_both_operator_string_and_string_view() == "success"
+

 def test_integer_casting():
    """Issue #929 - out-of-range integer values shouldn't be accepted"""
@ -299,7 +310,8 @@ def test_int_convert():
    assert noconvert(7) == 7
    cant_convert(3.14159)
    # TODO: Avoid DeprecationWarning in `PyLong_AsLong` (and similar)
-    if (3, 8) <= env.PY < (3, 10):
+    # TODO: PyPy 3.8 does not behave like CPython 3.8 here yet (7.3.7)
+    if (3, 8) <= env.PY < (3, 10) and env.CPYTHON:
        with env.deprecated_call():
            assert convert(Int()) == 42
    else:
@ -334,7 +346,9 @@ def test_numpy_int_convert():

    # The implicit conversion from np.float32 is undesirable but currently accepted.
    # TODO: Avoid DeprecationWarning in `PyLong_AsLong` (and similar)
-    if (3, 8) <= env.PY < (3, 10):
+    # TODO: PyPy 3.8 does not behave like CPython 3.8 here yet (7.3.7)
+    # https://github.com/pybind/pybind11/issues/3408
+    if (3, 8) <= env.PY < (3, 10) and env.CPYTHON:
        with env.deprecated_call():
            assert convert(np.float32(3.14159)) == 3
    else:
--- a/tests/test_copy_move.cpp
+++ b/tests/test_copy_move.cpp
@ -85,13 +85,13 @@ public:
 PYBIND11_NAMESPACE_BEGIN(pybind11)
 PYBIND11_NAMESPACE_BEGIN(detail)
 template <> struct type_caster<MoveOnlyInt> {
-    PYBIND11_TYPE_CASTER(MoveOnlyInt, _("MoveOnlyInt"));
+    PYBIND11_TYPE_CASTER(MoveOnlyInt, const_name("MoveOnlyInt"));
    bool load(handle src, bool) { value = MoveOnlyInt(src.cast<int>()); return true; }
    static handle cast(const MoveOnlyInt &m, return_value_policy r, handle p) { return pybind11::cast(m.value, r, p); }
 };

 template <> struct type_caster<MoveOrCopyInt> {
-    PYBIND11_TYPE_CASTER(MoveOrCopyInt, _("MoveOrCopyInt"));
+    PYBIND11_TYPE_CASTER(MoveOrCopyInt, const_name("MoveOrCopyInt"));
    bool load(handle src, bool) { value = MoveOrCopyInt(src.cast<int>()); return true; }
    static handle cast(const MoveOrCopyInt &m, return_value_policy r, handle p) { return pybind11::cast(m.value, r, p); }
 };
@ -100,7 +100,7 @@ template <> struct type_caster<CopyOnlyInt> {
 protected:
    CopyOnlyInt value;
 public:
-    static constexpr auto name = _("CopyOnlyInt");
+    static constexpr auto name = const_name("CopyOnlyInt");
    bool load(handle src, bool) { value = CopyOnlyInt(src.cast<int>()); return true; }
    static handle cast(const CopyOnlyInt &m, return_value_policy r, handle p) { return pybind11::cast(m.value, r, p); }
    static handle cast(const CopyOnlyInt *src, return_value_policy policy, handle parent) {
--- a/tests/test_custom_type_casters.cpp
+++ b/tests/test_custom_type_casters.cpp
@ -18,7 +18,12 @@ class ArgAlwaysConverts { };
 namespace pybind11 { namespace detail {
 template <> struct type_caster<ArgInspector1> {
 public:
+    // Classic
+#ifndef _
    PYBIND11_TYPE_CASTER(ArgInspector1, _("ArgInspector1"));
+#else
+    PYBIND11_TYPE_CASTER(ArgInspector1, const_name("ArgInspector1"));
+#endif

    bool load(handle src, bool convert) {
        value.arg = "loading ArgInspector1 argument " +
@ -33,7 +38,7 @@ public:
 };
 template <> struct type_caster<ArgInspector2> {
 public:
-    PYBIND11_TYPE_CASTER(ArgInspector2, _("ArgInspector2"));
+    PYBIND11_TYPE_CASTER(ArgInspector2, const_name("ArgInspector2"));

    bool load(handle src, bool convert) {
        value.arg = "loading ArgInspector2 argument " +
@ -48,7 +53,7 @@ public:
 };
 template <> struct type_caster<ArgAlwaysConverts> {
 public:
-    PYBIND11_TYPE_CASTER(ArgAlwaysConverts, _("ArgAlwaysConverts"));
+    PYBIND11_TYPE_CASTER(ArgAlwaysConverts, const_name("ArgAlwaysConverts"));

    bool load(handle, bool convert) {
        return convert;
@ -76,7 +81,7 @@ public:
 };
 namespace pybind11 { namespace detail {
 template <> struct type_caster<DestructionTester> {
-    PYBIND11_TYPE_CASTER(DestructionTester, _("DestructionTester"));
+    PYBIND11_TYPE_CASTER(DestructionTester, const_name("DestructionTester"));
    bool load(handle, bool) { return true; }

    static handle cast(const DestructionTester &, return_value_policy, handle) {
--- a/tests/test_embed/CMakeLists.txt
+++ b/tests/test_embed/CMakeLists.txt
@ -25,7 +25,7 @@ pybind11_enable_warnings(test_embed)
 target_link_libraries(test_embed PRIVATE pybind11::embed Catch2::Catch2 Threads::Threads)

 if(NOT CMAKE_CURRENT_SOURCE_DIR STREQUAL CMAKE_CURRENT_BINARY_DIR)
-  file(COPY test_interpreter.py DESTINATION "${CMAKE_CURRENT_BINARY_DIR}")
+  file(COPY test_interpreter.py test_trampoline.py DESTINATION "${CMAKE_CURRENT_BINARY_DIR}")
 endif()

 add_custom_target(
--- a/tests/test_embed/test_interpreter.cpp
+++ b/tests/test_embed/test_interpreter.cpp
@ -37,6 +37,22 @@ class PyWidget final : public Widget {
    std::string argv0() const override { PYBIND11_OVERRIDE_PURE(std::string, Widget, argv0); }
 };

+class test_override_cache_helper {
+
+public:
+    virtual int func() { return 0; }
+
+    test_override_cache_helper() = default;
+    virtual ~test_override_cache_helper() = default;
+    // Non-copyable
+    test_override_cache_helper &operator=(test_override_cache_helper const &Right) = delete;
+    test_override_cache_helper(test_override_cache_helper const &Copy) = delete;
+};
+
+class test_override_cache_helper_trampoline : public test_override_cache_helper {
+    int func() override { PYBIND11_OVERRIDE(int, test_override_cache_helper, func); }
+};
+
 PYBIND11_EMBEDDED_MODULE(widget_module, m) {
    py::class_<Widget, PyWidget>(m, "Widget")
        .def(py::init<std::string>())
@ -45,6 +61,12 @@ PYBIND11_EMBEDDED_MODULE(widget_module, m) {
    m.def("add", [](int i, int j) { return i + j; });
 }

+PYBIND11_EMBEDDED_MODULE(trampoline_module, m) {
+    py::class_<test_override_cache_helper, test_override_cache_helper_trampoline, std::shared_ptr<test_override_cache_helper>>(m, "test_override_cache_helper")
+        .def(py::init_alias<>())
+        .def("func", &test_override_cache_helper::func);
+}
+
 PYBIND11_EMBEDDED_MODULE(throw_exception, ) {
    throw std::runtime_error("C++ Error");
 }
@ -73,6 +95,33 @@ TEST_CASE("Pass classes and data between modules defined in C++ and Python") {
    REQUIRE(cpp_widget.the_answer() == 42);
 }

+TEST_CASE("Override cache") {
+    auto module_ = py::module_::import("test_trampoline");
+    REQUIRE(py::hasattr(module_, "func"));
+    REQUIRE(py::hasattr(module_, "func2"));
+
+    auto locals = py::dict(**module_.attr("__dict__"));
+
+    int i = 0;
+    for (; i < 1500; ++i) {
+        std::shared_ptr<test_override_cache_helper> p_obj;
+        std::shared_ptr<test_override_cache_helper> p_obj2;
+
+        py::object loc_inst = locals["func"]();
+        p_obj = py::cast<std::shared_ptr<test_override_cache_helper>>(loc_inst);
+
+        int ret = p_obj->func();
+
+        REQUIRE(ret == 42);
+
+        loc_inst = locals["func2"]();
+
+        p_obj2 = py::cast<std::shared_ptr<test_override_cache_helper>>(loc_inst);
+
+        p_obj2->func();
+    }
+}
+
 TEST_CASE("Import error handling") {
    REQUIRE_NOTHROW(py::module_::import("widget_module"));
    REQUIRE_THROWS_WITH(py::module_::import("throw_exception"),
--- a/tests/test_embed/test_trampoline.py
+++ b/tests/test_embed/test_trampoline.py
@ -0,0 +1,18 @@
+# -*- coding: utf-8 -*-
+
+import trampoline_module
+
+
+def func():
+    class Test(trampoline_module.test_override_cache_helper):
+        def func(self):
+            return 42
+
+    return Test()
+
+
+def func2():
+    class Test(trampoline_module.test_override_cache_helper):
+        pass
+
+    return Test()
--- a/tests/test_exceptions.py
+++ b/tests/test_exceptions.py
@ -3,7 +3,7 @@ import sys

 import pytest

-import env  # noqa: F401
+import env
 import pybind11_cross_module_tests as cm
 from pybind11_tests import exceptions as m

@ -97,6 +97,8 @@ def ignore_pytest_unraisable_warning(f):
        return f


+# TODO: find out why this fails on PyPy, https://foss.heptapod.net/pypy/pypy/-/issues/3583
+@pytest.mark.xfail(env.PYPY, reason="Failure on PyPy 3.8 (7.3.7)", strict=False)
@ignore_pytest_unraisable_warning
 def test_python_alreadyset_in_destructor(monkeypatch, capsys):
    hooked = False
--- a/tests/test_kwargs_and_defaults.cpp
+++ b/tests/test_kwargs_and_defaults.cpp
@ -56,6 +56,23 @@ TEST_SUBMODULE(kwargs_and_defaults, m) {
    m.def("mixed_plus_args_kwargs_defaults", mixed_plus_both,
            py::arg("i") = 1, py::arg("j") = 3.14159);

+    m.def("args_kwonly",
+            [](int i, double j, const py::args &args, int z) { return py::make_tuple(i, j, args, z); },
+            "i"_a, "j"_a, "z"_a);
+    m.def("args_kwonly_kwargs",
+            [](int i, double j, const py::args &args, int z, const py::kwargs &kwargs) {
+                return py::make_tuple(i, j, args, z, kwargs); },
+            "i"_a, "j"_a, py::kw_only{}, "z"_a);
+    m.def("args_kwonly_kwargs_defaults",
+            [](int i, double j, const py::args &args, int z, const py::kwargs &kwargs) {
+                return py::make_tuple(i, j, args, z, kwargs); },
+            "i"_a = 1, "j"_a = 3.14159, "z"_a = 42);
+    m.def("args_kwonly_full_monty",
+            [](int h, int i, double j, const py::args &args, int z, const py::kwargs &kwargs) {
+                return py::make_tuple(h, i, j, args, z, kwargs); },
+            py::arg() = 1, py::arg() = 2, py::pos_only{}, "j"_a = 3.14159, "z"_a = 42);
+
+
    // test_args_refcount
    // PyPy needs a garbage collection to get the reference count values to match CPython's behaviour
    #ifdef PYPY_VERSION
@ -150,4 +167,21 @@ TEST_SUBMODULE(kwargs_and_defaults, m) {
        "class_default_argument",
        [](py::object a) { return py::repr(std::move(a)); },
        "a"_a = py::module_::import("decimal").attr("Decimal"));
+
+    // Initial implementation of kw_only was broken when used on a method/constructor before any
+    // other arguments
+    // https://github.com/pybind/pybind11/pull/3402#issuecomment-963341987
+
+    struct first_arg_kw_only {};
+    py::class_<first_arg_kw_only>(m, "first_arg_kw_only")
+        .def(py::init([](int) { return first_arg_kw_only(); }),
+             py::kw_only(), // This being before any args was broken
+             py::arg("i") = 0)
+        .def("method", [](first_arg_kw_only&, int, int) {},
+             py::kw_only(), // and likewise here
+             py::arg("i") = 1, py::arg("j") = 2)
+        // Closely related: pos_only marker didn't show up properly when it was before any other
+        // arguments (although that is fairly useless in practice).
+        .def("pos_only", [](first_arg_kw_only&, int, int) {},
+                py::pos_only{}, py::arg("i"), py::arg("j"));
 }
--- a/tests/test_kwargs_and_defaults.py
+++ b/tests/test_kwargs_and_defaults.py
@ -141,6 +141,53 @@ def test_mixed_args_and_kwargs(msg):
    """  # noqa: E501 line too long
    )

+    # Arguments after a py::args are automatically keyword-only (pybind 2.9+)
+    assert m.args_kwonly(2, 2.5, z=22) == (2, 2.5, (), 22)
+    assert m.args_kwonly(2, 2.5, "a", "b", "c", z=22) == (2, 2.5, ("a", "b", "c"), 22)
+    assert m.args_kwonly(z=22, i=4, j=16) == (4, 16, (), 22)
+
+    with pytest.raises(TypeError) as excinfo:
+        assert m.args_kwonly(2, 2.5, 22)  # missing z= keyword
+    assert (
+        msg(excinfo.value)
+        == """
+        args_kwonly(): incompatible function arguments. The following argument types are supported:
+            1. (i: int, j: float, *args, z: int) -> tuple
+
+        Invoked with: 2, 2.5, 22
+    """
+    )
+
+    assert m.args_kwonly_kwargs(i=1, k=4, j=10, z=-1, y=9) == (
+        1,
+        10,
+        (),
+        -1,
+        {"k": 4, "y": 9},
+    )
+    assert m.args_kwonly_kwargs(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, z=11, y=12) == (
+        1,
+        2,
+        (3, 4, 5, 6, 7, 8, 9, 10),
+        11,
+        {"y": 12},
+    )
+    assert (
+        m.args_kwonly_kwargs.__doc__
+        == "args_kwonly_kwargs(i: int, j: float, *args, z: int, **kwargs) -> tuple\n"
+    )
+
+    assert (
+        m.args_kwonly_kwargs_defaults.__doc__
+        == "args_kwonly_kwargs_defaults(i: int = 1, j: float = 3.14159, *args, z: int = 42, **kwargs) -> tuple\n"  # noqa: E501 line too long
+    )
+    assert m.args_kwonly_kwargs_defaults() == (1, 3.14159, (), 42, {})
+    assert m.args_kwonly_kwargs_defaults(2) == (2, 3.14159, (), 42, {})
+    assert m.args_kwonly_kwargs_defaults(z=-99) == (1, 3.14159, (), -99, {})
+    assert m.args_kwonly_kwargs_defaults(5, 6, 7, 8) == (5, 6, (7, 8), 42, {})
+    assert m.args_kwonly_kwargs_defaults(5, 6, 7, m=8) == (5, 6, (7,), 42, {"m": 8})
+    assert m.args_kwonly_kwargs_defaults(5, 6, 7, m=8, z=9) == (5, 6, (7,), 9, {"m": 8})
+

 def test_keyword_only_args(msg):
    assert m.kw_only_all(i=1, j=2) == (1, 2)
@ -178,10 +225,23 @@ def test_keyword_only_args(msg):
    assert (
        msg(excinfo.value)
        == """
-        arg(): cannot specify an unnamed argument after an kw_only() annotation
+        arg(): cannot specify an unnamed argument after a kw_only() annotation or args() argument
    """
    )

+    # https://github.com/pybind/pybind11/pull/3402#issuecomment-963341987
+    x = m.first_arg_kw_only(i=1)
+    x.method()
+    x.method(i=1, j=2)
+    assert (
+        m.first_arg_kw_only.__init__.__doc__
+        == "__init__(self: pybind11_tests.kwargs_and_defaults.first_arg_kw_only, *, i: int = 0) -> None\n"  # noqa: E501 line too long
+    )
+    assert (
+        m.first_arg_kw_only.method.__doc__
+        == "method(self: pybind11_tests.kwargs_and_defaults.first_arg_kw_only, *, i: int = 1, j: int = 2) -> None\n"  # noqa: E501 line too long
+    )
+

 def test_positional_only_args(msg):
    assert m.pos_only_all(1, 2) == (1, 2)
@ -222,6 +282,55 @@ def test_positional_only_args(msg):
        m.pos_only_def_mix(1, j=4)
    assert "incompatible function arguments" in str(excinfo.value)

+    # Mix it with args and kwargs:
+    assert (
+        m.args_kwonly_full_monty.__doc__
+        == "args_kwonly_full_monty(arg0: int = 1, arg1: int = 2, /, j: float = 3.14159, *args, z: int = 42, **kwargs) -> tuple\n"  # noqa: E501 line too long
+    )
+    assert m.args_kwonly_full_monty() == (1, 2, 3.14159, (), 42, {})
+    assert m.args_kwonly_full_monty(8) == (8, 2, 3.14159, (), 42, {})
+    assert m.args_kwonly_full_monty(8, 9) == (8, 9, 3.14159, (), 42, {})
+    assert m.args_kwonly_full_monty(8, 9, 10) == (8, 9, 10.0, (), 42, {})
+    assert m.args_kwonly_full_monty(3, 4, 5, 6, 7, m=8, z=9) == (
+        3,
+        4,
+        5.0,
+        (
+            6,
+            7,
+        ),
+        9,
+        {"m": 8},
+    )
+    assert m.args_kwonly_full_monty(3, 4, 5, 6, 7, m=8, z=9) == (
+        3,
+        4,
+        5.0,
+        (
+            6,
+            7,
+        ),
+        9,
+        {"m": 8},
+    )
+    assert m.args_kwonly_full_monty(5, j=7, m=8, z=9) == (5, 2, 7.0, (), 9, {"m": 8})
+    assert m.args_kwonly_full_monty(i=5, j=7, m=8, z=9) == (
+        1,
+        2,
+        7.0,
+        (),
+        9,
+        {"i": 5, "m": 8},
+    )
+
+    # pos_only at the beginning of the argument list was "broken" in how it was displayed (though
+    # this is fairly useless in practice).  Related to:
+    # https://github.com/pybind/pybind11/pull/3402#issuecomment-963341987
+    assert (
+        m.first_arg_kw_only.pos_only.__doc__
+        == "pos_only(self: pybind11_tests.kwargs_and_defaults.first_arg_kw_only, /, i: int, j: int) -> None\n"  # noqa: E501 line too long
+    )
+

 def test_signatures():
    assert "kw_only_all(*, i: int, j: int) -> tuple\n" == m.kw_only_all.__doc__
--- a/tests/test_methods_and_attributes.cpp
+++ b/tests/test_methods_and_attributes.cpp
@ -159,6 +159,14 @@ struct RefQualified {
    int constRefQualified(int other) const & { return value + other; }
 };

+// Test rvalue ref param
+struct RValueRefParam {
+    std::size_t func1(std::string&& s) { return s.size(); }
+    std::size_t func2(std::string&& s) const { return s.size(); }
+    std::size_t func3(std::string&& s) & { return s.size(); }
+    std::size_t func4(std::string&& s) const & { return s.size(); }
+};
+
 TEST_SUBMODULE(methods_and_attributes, m) {
    // test_methods_and_attributes
    py::class_<ExampleMandA> emna(m, "ExampleMandA");
@ -409,4 +417,11 @@ TEST_SUBMODULE(methods_and_attributes, m) {
        .def_readonly("value", &RefQualified::value)
        .def("refQualified", &RefQualified::refQualified)
        .def("constRefQualified", &RefQualified::constRefQualified);
+
+    py::class_<RValueRefParam>(m, "RValueRefParam")
+        .def(py::init<>())
+        .def("func1", &RValueRefParam::func1)
+        .def("func2", &RValueRefParam::func2)
+        .def("func3", &RValueRefParam::func3)
+        .def("func4", &RValueRefParam::func4);
 }
--- a/tests/test_methods_and_attributes.py
+++ b/tests/test_methods_and_attributes.py
@ -515,3 +515,11 @@ def test_overload_ordering():
    assert "2. (arg0: {}) -> int".format(uni_name) in str(err.value)
    assert "3. (arg0: {}) -> int".format(uni_name) in str(err.value)
    assert "4. (arg0: int) -> int" in str(err.value)
+
+
+def test_rvalue_ref_param():
+    r = m.RValueRefParam()
+    assert r.func1("123") == 3
+    assert r.func2("1234") == 4
+    assert r.func3("12345") == 5
+    assert r.func4("123456") == 6
--- a/tests/test_virtual_functions.cpp
+++ b/tests/test_virtual_functions.cpp
@ -214,6 +214,25 @@ static void test_gil_from_thread() {
    t.join();
 }

+class test_override_cache_helper {
+
+public:
+    virtual int func() { return 0; }
+
+    test_override_cache_helper() = default;
+    virtual ~test_override_cache_helper() = default;
+    // Non-copyable
+    test_override_cache_helper &operator=(test_override_cache_helper const &Right) = delete;
+    test_override_cache_helper(test_override_cache_helper const &Copy) = delete;
+};
+
+class test_override_cache_helper_trampoline : public test_override_cache_helper {
+    int func() override { PYBIND11_OVERRIDE(int, test_override_cache_helper, func); }
+};
+
+inline int test_override_cache(std::shared_ptr<test_override_cache_helper> const &instance) { return instance->func(); }
+
+

 // Forward declaration (so that we can put the main tests here; the inherited virtual approaches are
 // rather long).
@ -378,6 +397,12 @@ TEST_SUBMODULE(virtual_functions, m) {
 //      .def("str_ref", &OverrideTest::str_ref)
        .def("A_value", &OverrideTest::A_value)
        .def("A_ref", &OverrideTest::A_ref);
+
+    py::class_<test_override_cache_helper, test_override_cache_helper_trampoline, std::shared_ptr<test_override_cache_helper>>(m, "test_override_cache_helper")
+        .def(py::init_alias<>())
+        .def("func", &test_override_cache_helper::func);
+
+    m.def("test_override_cache", test_override_cache);
 }


--- a/tests/test_virtual_functions.py
+++ b/tests/test_virtual_functions.py
@ -439,3 +439,22 @@ def test_issue_1454():
    # Fix issue #1454 (crash when acquiring/releasing GIL on another thread in Python 2.7)
    m.test_gil()
    m.test_gil_from_thread()
+
+
+def test_python_override():
+    def func():
+        class Test(m.test_override_cache_helper):
+            def func(self):
+                return 42
+
+        return Test()
+
+    def func2():
+        class Test(m.test_override_cache_helper):
+            pass
+
+        return Test()
+
+    for _ in range(1500):
+        assert m.test_override_cache(func()) == 42
+        assert m.test_override_cache(func2()) == 0
--- a/tools/pybind11Common.cmake
+++ b/tools/pybind11Common.cmake
@ -318,13 +318,21 @@ function(_pybind11_generate_lto target prefer_thin_lto)
      set(cxx_append ";-fno-fat-lto-objects")
    endif()

-    if(CMAKE_CXX_COMPILER_ID MATCHES "Clang" AND prefer_thin_lto)
+    if(CMAKE_SYSTEM_PROCESSOR MATCHES "ppc64le" OR CMAKE_SYSTEM_PROCESSOR MATCHES "mips64")
+      set(NO_FLTO_ARCH TRUE)
+    else()
+      set(NO_FLTO_ARCH FALSE)
+    endif()
+
+    if(CMAKE_CXX_COMPILER_ID MATCHES "Clang"
+       AND prefer_thin_lto
+       AND NOT NO_FLTO_ARCH)
      _pybind11_return_if_cxx_and_linker_flags_work(
        HAS_FLTO_THIN "-flto=thin${cxx_append}" "-flto=thin${linker_append}"
        PYBIND11_LTO_CXX_FLAGS PYBIND11_LTO_LINKER_FLAGS)
    endif()

-    if(NOT HAS_FLTO_THIN)
+    if(NOT HAS_FLTO_THIN AND NOT NO_FLTO_ARCH)
      _pybind11_return_if_cxx_and_linker_flags_work(
        HAS_FLTO "-flto${cxx_append}" "-flto${linker_append}" PYBIND11_LTO_CXX_FLAGS
        PYBIND11_LTO_LINKER_FLAGS)
--- a/tools/pybind11Config.cmake.in
+++ b/tools/pybind11Config.cmake.in
@ -13,7 +13,7 @@ This module sets the following variables in your project:
 ``pybind11_VERSION``
  pybind11 version in format Major.Minor.Release
 ``pybind11_VERSION_TYPE``
-  pybind11 version type (dev, release)
+  pybind11 version type (``dev*`` or empty for a release)
 ``pybind11_INCLUDE_DIRS``
  Directories where pybind11 and python headers are located.
 ``pybind11_INCLUDE_DIR``
@ -228,6 +228,6 @@ include("${CMAKE_CURRENT_LIST_DIR}/pybind11Common.cmake")
 if(NOT pybind11_FIND_QUIETLY)
  message(
    STATUS
-      "Found pybind11: ${pybind11_INCLUDE_DIR} (found version \"${pybind11_VERSION}\" ${pybind11_VERSION_TYPE})"
+      "Found pybind11: ${pybind11_INCLUDE_DIR} (found version \"${pybind11_VERSION}${pybind11_VERSION_TYPE}\")"
  )
 endif()