使用matplotlibcpp.h与plot()函数时出现Bus error问题（macOS环境）

Hey there, sorry you're hitting this bus error while trying out C++ plotting with matplotlibcpp—let's break this down and get you up and running!

First, let's understand why this bus error might be happening #

On macOS, bus errors often pop up when there's a mismatch between the Python your code is trying to link to and the one you actually have installed (Homebrew's Python 3.9, in your case). Since your "Hello World" code works, the issue definitely lies in how matplotlibcpp is interacting with your Python environment during compilation/runtime.

Step 1: Fix your compilation command #

The default g++ call probably isn't pointing to your Homebrew Python 3.9 correctly. You need to explicitly tell the compiler where to find Python's headers and libraries.

1. First, get your Homebrew Python 3.9 paths:

   • Run this in terminal to get the header directory:

     python3.9 -c "import sysconfig; print(sysconfig.get_path('include'))"
     

     It'll output something like /usr/local/Cellar/python@3.9/3.9.16/Frameworks/Python.framework/Versions/3.9/include/python3.9
   • Run this for the library directory:

     python3.9 -c "import sysconfig; print(sysconfig.get_config_var('LIBDIR'))"
     

     Output example: /usr/local/Cellar/python@3.9/3.9.16/Frameworks/Python.framework/Versions/3.9/lib

2. Use these paths in your compilation command:
   Replace the placeholders with your actual paths:

   g++-11 -std=c++17 your_code.cpp -o plotter -I[YOUR_HEADER_PATH] -L[YOUR_LIBRARY_PATH] -lpython3.9 -framework Python
   

   • -std=c++17: Ensures compatibility with modern C++ features matplotlibcpp needs
   • -I: Tells the compiler where to find Python headers
   • -L: Points to Python's library files
   • -lpython3.9: Links against the Python 3.9 library
   • -framework Python: A macOS-specific flag required for Python integration

Step 2: Tweak your code slightly #

Adding a file extension to savefig can avoid format-related issues. Update your code like this:

#include <vector>
#include "matplotlibcpp.h"
int main() {
    std::vector<int> x{1,2,3,4};
    std::vector<int> y{1,2,3,4};
    matplotlibcpp::plot(x, y);
    matplotlibcpp::savefig("figure.png"); // Add .png extension here
    return 0;
}

Step 3: Verify your Python dependencies #

Make sure matplotlib is installed for your Homebrew Python 3.9:

pip3.9 install matplotlib

Avoid using plain pip here—it might target the system's default Python instead.

Step 4: Set up VS Code for easy compilation (optional) #

If you're using VS Code's build tasks, update your .vscode/tasks.json with the correct compilation parameters:

{
    "version": "2.0.0",
    "tasks": [
        {
            "label": "Compile Plot Code",
            "type": "shell",
            "command": "g++-11 -std=c++17 ${file} -o ${fileBasenameNoExtension} -I[YOUR_HEADER_PATH] -L[YOUR_LIBRARY_PATH] -lpython3.9 -framework Python",
            "group": {
                "kind": "build",
                "isDefault": true
            },
            "problemMatcher": ["$gcc"]
        }
    ]
}

Extra troubleshooting tip #

If you still hit errors, try setting the PYTHONHOME environment variable before running your compiled program:

export PYTHONHOME=/usr/local/Cellar/python@3.9/3.9.16/Frameworks/Python.framework/Versions/3.9
./plotter

This ensures your program uses the correct Python runtime.

内容的提问来源于stack exchange，提问作者Julia