[msic] Increase numpy requirement.

.github/workflows/linux.yml

@@ -66,7 +66,7 @@ jobs:
         sudo apt install -y gdb gnupg curl wget build-essential cmake doxygen graphviz texlive-latex-base
 
         "${PYTHON_EXECUTABLE}" -m pip install setuptools wheel "pip>=20.3"
-        "${PYTHON_EXECUTABLE}" -m pip install "numpy>=1.21,<2.0"
+        "${PYTHON_EXECUTABLE}" -m pip install "numpy>=1.23,<2.0"
 
         git config --global advice.detachedHead false
     - name: Build jiminy dependencies

.github/workflows/macos.yml

@@ -68,7 +68,7 @@ jobs:
         echo "InstallDir=${GITHUB_WORKSPACE}/install" >> $GITHUB_ENV
 
         "${PYTHON_EXECUTABLE}" -m pip install setuptools wheel "pip>=20.3"
-        "${PYTHON_EXECUTABLE}" -m pip install "numpy>=1.21,<2.0"
+        "${PYTHON_EXECUTABLE}" -m pip install "numpy>=1.23,<2.0"
         "${PYTHON_EXECUTABLE}" -m pip install delocate twine
     - name: Build jiminy dependencies
       run: |

.github/workflows/manylinux.yml

@@ -62,7 +62,7 @@ jobs:
         echo "InstallDir=${GITHUB_WORKSPACE}/install" >> $GITHUB_ENV
 
         "${PYTHON_EXECUTABLE}" -m pip install setuptools wheel "pip>=20.3"
-        "${PYTHON_EXECUTABLE}" -m pip install "numpy>=1.21,<2.0"
+        "${PYTHON_EXECUTABLE}" -m pip install "numpy>=1.23,<2.0"
         "${PYTHON_EXECUTABLE}" -m pip install twine cmake
     - name: Build project dependencies
       run: |
@@ -78,7 +78,8 @@ jobs:
         mkdir "$RootDir/build"
         cd "$RootDir/build"
         export LD_LIBRARY_PATH="$InstallDir/lib:$InstallDir/lib64:/usr/local/lib"
-        cmake "$RootDir" -DCMAKE_INSTALL_PREFIX="$InstallDir" -DCMAKE_PREFIX_PATH="$InstallDir" \
+        cmake "$RootDir" -Wdev \
+              -DCMAKE_INSTALL_PREFIX="$InstallDir" -DCMAKE_PREFIX_PATH="$InstallDir" \
               -DCMAKE_INTERPROCEDURAL_OPTIMIZATION=ON -DCMAKE_VERBOSE_MAKEFILE=ON \
               -DBOOST_ROOT="$InstallDir" -DBoost_INCLUDE_DIR="$InstallDir/include" \
               -DBoost_NO_SYSTEM_PATHS=TRUE -DBoost_NO_BOOST_CMAKE=TRUE \

.github/workflows/win.yml

@@ -56,7 +56,7 @@ jobs:
       run: |
         git config --global advice.detachedHead false
         python -m pip install setuptools wheel "pip>=20.3"
-        python -m pip install "numpy>=1.21,<2.0"
+        python -m pip install "numpy>=1.23,<2.0"
         python -m pip install pefile machomachomangler
     - name: Build jiminy dependencies
       run: |

INSTALL.md

@@ -50,7 +50,7 @@ python -m pip install --prefer-binary gym-jiminy[all]
 
 ## Excluding dependencies on Ubuntu 18+
 
-First, one must install the pre-compiled libraries of the dependencies. Most of them are available on `robotpkg` APT repository. Just run the bash script to install them automatically for Ubuntu 18 and upward. It should be straightforward to adapt it to any other distribution for which `robotpkg` is available. Note that if you plan to use a virtual environment (`venv`, `pyenv`, ...), you must install the Python dependencies manually using `pip`, i.e. `"wheel", "numpy>=1.21,<2.0"`.
+First, one must install the pre-compiled libraries of the dependencies. Most of them are available on `robotpkg` APT repository. Just run the bash script to install them automatically for Ubuntu 18 and upward. It should be straightforward to adapt it to any other distribution for which `robotpkg` is available. Note that if you plan to use a virtual environment (`venv`, `pyenv`, ...), you must install the Python dependencies manually using `pip`, i.e. `"wheel", "numpy>=1.23,<2.0"`.
 
 ```bash
 sudo env "PATH=$PATH" ./build_tools/easy_install_deps_ubuntu.sh
@@ -79,7 +79,7 @@ make install -j2
 
 ```bash
 sudo apt install -y gnupg curl wget build-essential cmake doxygen graphviz
-python -m pip install "wheel", "numpy>=1.21,<2.0"
+python -m pip install "wheel", "numpy>=1.23,<2.0"
 ```
 
 ### Jiminy dependencies build and install
@@ -121,7 +121,7 @@ You have to preinstall by yourself the (free) MSVC 2019 toolchain.
 Then, install `setuptools`, `wheel` and `numpy`.
 
 ```powershell
-python -m pip install setuptools wheel "numpy>=1.21,<2.0"
+python -m pip install setuptools wheel "numpy>=1.23,<2.0"
 ```
 
 ### Jiminy dependencies build and install

build_tools/build_install_deps_unix.sh

@@ -271,7 +271,7 @@ else
   echo "Build type '${BUILD_TYPE}' not supported." >&2
   exit 1
 fi
-CMAKE_CXX_FLAGS_B2="-fPIC -std=c++11"
+CMAKE_CXX_FLAGS_B2="-fPIC -std=c++17"
 if [ "${OSTYPE//[0-9.]/}" == "darwin" ]; then
   CMAKE_CXX_FLAGS_B2="${CMAKE_CXX_FLAGS_B2} -mmacosx-version-min=${OSX_DEPLOYMENT_TARGET}"
 fi

build_tools/easy_install_deps_ubuntu.sh

@@ -59,7 +59,7 @@ echo "-- Python writable site-packages: ${PYTHON_SITELIB}"
 apt update && \
 apt install -y python3-pip && \
 ${SUDO_CMD} python3 -m pip install setuptools wheel "pip>=20.3" && \
-${SUDO_CMD} python3 -m pip install "numpy>=1.21,<2.0" "numba>=0.54.0"
+${SUDO_CMD} python3 -m pip install "numpy>=1.23,<2.0" "numba>=0.54.0"
 
 # Install standard linux utilities
 apt install -y gnupg curl wget build-essential cmake doxygen graphviz pandoc

python/jiminy_py/setup.py

@@ -85,19 +85,16 @@ def finalize_options(self) -> None:
         "jiminy_replay=jiminy_py.viewer.replay:_play_logs_files_entrypoint"
     ]},
     install_requires=[
-        # Display elegant and versatile process bar.
-        "tqdm",
         # Standard library for matrix algebra.
         # - 1.20 breaks ABI
         # - >=1.21,<1.21.5 is causing segfault with boost::python.
         #   See issue: https://github.com/boostorg/python/issues/376
         # - 1.22 breaks API for compiled libs.
-        np_req,
+        # - 2.0 is backward compatible up to 1.23, but not forward compatible.
+        #   see: https://numpy.org/devdocs/dev/depending_on_numpy.html
+        "numpy>=1.23,<2.0",
         # Parser for Jiminy's hardware description file.
         "toml",
-        # Used internally by Robot to replace meshes by associated minimal
-        # volume bounding box.
-        "trimesh",
         # Standalone cross-platform mesh visualizer used as Viewer's backend.
         # Panda3d is NOT supported by PyPy even if built from source.
         # - 1.10.12 fixes numerous bugs
@@ -111,7 +108,12 @@ def finalize_options(self) -> None:
         # Used internally by Viewer to record video programmatically when
         # Panda3d is used as rendering backend.
         # - >= 8.0.0 provides cross-platform precompiled binary wheels
-        "av>=8.0.0"
+        "av>=8.0.0",
+        # Used internally by Robot to replace meshes by associated minimal
+        # volume bounding box.
+        "trimesh",
+        # Display elegant and versatile process bar.
+        "tqdm"
     ],
     extras_require={
         "plot": [