Document compiler setup and fix conda-build channel requirements

1. Added detailed documentation for conda users building with pip:
   - Need to install gxx_linux-64 and gcc_linux-64 compilers
   - Prevents GLIBCXX version mismatch errors

2. Fixed conda-build documentation:
   - Added -c conda-forge flag requirement
   - Explains this is needed for cmake, ninja, and compiler dependencies

3. Enhanced troubleshooting section:
   - Compiler issues in conda environments
   - Conda build packages not found error
   - Clear solutions for both pip and conda-build workflows

Co-authored-by: Ruochun <24469442+Ruochun@users.noreply.github.com>

Author: Copilot

PYTHON_BUILD.md

@@ -16,6 +16,23 @@ This document describes how to build the Python package for DEM-Engine.
 
 ### Option 1: Using pip (recommended)
 
+**Important for Conda Users:** If you're building in a conda environment, you need to install conda's compilers first to ensure compatibility:
+
+```bash
+# Activate your conda environment
+conda activate your_env_name
+
+# Install conda's compilers
+conda install -c conda-forge gxx_linux-64 gcc_linux-64
+
+# Then build/install
+pip install .
+```
+
+Without conda's compilers, the wheel will link against your system's libstdc++ which may cause import errors in conda environments.
+
+For non-conda environments:
+
 ```bash
 pip install .
 ```
@@ -24,6 +41,8 @@ This will use scikit-build-core to automatically configure CMake with the Python
 
 ### Option 2: Building a wheel
 
+**Important for Conda Users:** If building in a conda environment, install conda's compilers first (see Option 1).
+
 ```bash
 pip install build
 python -m build
@@ -41,16 +60,21 @@ pip install -e .
 
 ### Option 4: Using conda
 
-For conda users, you can build and install using conda-build:
+For conda users, you can build and install using conda-build. This method automatically handles all dependencies including compilers:
 
 ```bash
-# Build the conda package
-conda build recipe/
+# Build the conda package (requires conda-build and access to conda-forge channel)
+conda install conda-build
+conda build recipe/ -c conda-forge
 
 # Install from local build
 conda install --use-local deme
 ```
 
+**Note:** The `-c conda-forge` flag is required to access build dependencies like cmake, ninja, and the C/C++ compilers.
+conda install --use-local deme
+```
+
 This is useful for managing the package alongside other conda packages in your environment.
 
 ## CMake Options
@@ -96,6 +120,29 @@ The Python package provides bindings for:
 
 ## Troubleshooting
 
+### Compiler Issues in Conda Environments
+
+**Problem:** When building with `pip install .` in a conda environment, you get import errors like `GLIBCXX_3.4.30 not found`.
+
+**Solution:** Install conda's compilers before building:
+```bash
+conda install -c conda-forge gxx_linux-64 gcc_linux-64
+pip install .
+```
+
+This ensures the wheel is built with libraries compatible with your conda environment.
+
+### Conda Build: Packages Not Found Error
+
+**Problem:** `conda build recipe/` fails with "PackagesNotFoundError: cmake not available from current channels"
+
+**Solution:** Specify the conda-forge channel when building:
+```bash
+conda build recipe/ -c conda-forge
+```
+
+The conda-forge channel provides cmake, ninja, and other build dependencies.
+
 ### CUDA Not Found
 
 Ensure that CUDA is installed and the `CUDA_ROOT` or `CUDACXX` environment variable is set:

recipe/meta.yaml

@@ -10,6 +10,9 @@ build:
   script: {{ PYTHON }} -m pip install . -vvv
   skip: true  # [not linux]
 
+# Note: Build with: conda build recipe/ -c conda-forge
+# The conda-forge channel is required for build dependencies
+
 requirements:
   build:
     - {{ compiler('c') }}