Acquiring and Configuring Python 3.6 in Anaconda: A Comprehensive Guide from Historical Versions to Environment Management

Problem Context and Requirements Analysis

In deep learning project development, particularly when using TensorFlow for object detection, Python version compatibility often presents technical challenges. Users encountering issues while running object detection demos with Anaconda 3 and Python 3.7 have found that switching to Python 3.6 may resolve these compatibility problems. However, the official Anaconda download page only offers Python 3.7 and 2.7 versions, necessitating exploration of alternative approaches to obtain Python 3.6 environments.

Solution 1: Python Downgrade via conda

For users with the latest Anaconda installation, the most direct solution is to downgrade the Python version through the conda package manager. Execute the following command to switch the current environment to Python 3.6.0:

conda install python=3.6.0

This method is straightforward but carries potential risks. The downgrade process may cause dependency conflicts among installed packages, especially when certain packages require specific Python versions. While testing hasn't revealed major issues, actual outcomes depend on the specific package configuration in the user's environment. It's recommended to back up the current environment state before proceeding.

Solution 2: Downloading Historical Anaconda Versions

To avoid compatibility issues from environment downgrades, users can directly download historical Anaconda versions containing Python 3.6. Anaconda maintains all historical installation packages in its official archive repository.

Key version information includes:

• Starting with Anaconda 5.3.0 (released September 28, 2018), installation packages default to Python 3.7
• The last version supporting Python 3.6 is Anaconda 5.2.0 (released May 30, 2018)

Users can select appropriate installation packages based on their operating system:

# Linux 64-bit systems
Anaconda3-5.2.0-Linux-x86_64.sh

# Windows 64-bit systems
Anaconda3-5.2.0-Windows-x86_64.exe

# macOS systems
Anaconda3-5.2.0-MacOSX-x86_64.pkg

To verify the Python version included in a specific Anaconda release, consult the official release notes, looking for version change records in the format "python A.B.C -> X.Y.Z".

Solution 3: conda Environment Management

For users wishing to maintain their current Anaconda installation while using Python 3.6, conda environment management offers the most flexible solution. Creating isolated environments prevents interference with the main environment configuration.

The complete workflow for creating a Python 3.6 environment is as follows:

# Create a new environment named tensorflow-36 with Python 3.6
conda create -n tensorflow-36 python=3.6

# Activate the newly created environment
conda activate tensorflow-36

# Install TensorFlow and other necessary packages in the new environment
conda install tensorflow

The advantage of environment management lies in its isolation capability. Each environment can have independent package configurations and Python versions, making it particularly suitable for scenarios requiring maintenance of multiple projects or testing different version compatibilities. Environment switching commands are simple and intuitive, with no impact on other system components.

Solution Comparison and Selection Recommendations

Each solution suits different scenarios:

1. conda downgrade: Suitable for quick testing but carries higher risks and may compromise existing environment stability
2. Historical version installation: Provides the cleanest Python 3.6 environment, ideal for new projects or fresh installations
3. Environment management: The most recommended approach, balancing flexibility and stability while supporting parallel version usage

For TensorFlow object detection projects, considering deep learning frameworks' sensitivity to version dependencies, environment management is recommended as the primary approach. This ensures project environment reproducibility and isolation, facilitating team collaboration and deployment.

Technical Details and Best Practices

When implementing these solutions, pay attention to the following technical details:

1. Version precision: Python 3.6 includes multiple subversions (e.g., 3.6.0, 3.6.5), and certain packages may have specific requirements. conda commands support precise version specification:

conda create -n myenv python=3.6.5

2. Environment export and sharing: With conda environment management, export environment configurations using:

conda env export > environment.yml

The generated YAML file contains all packages with their exact versions, enabling environment reproduction on other machines.

3. Package compatibility checking: Before downgrading Python or installing historical versions, use conda's dry-run feature to check potential conflicts:

conda install python=3.6.0 --dry-run

4. System path configuration: When using environment management, ensure proper configuration of system PATH variables to avoid command conflicts between different environments.

Conclusion

While obtaining Python 3.6 environments in Anaconda isn't directly available through the official download page, users still have multiple reliable options through conda's flexibility and historical version archives. Environment management emerges as the best practice for most scenarios due to its superior isolation and flexibility, particularly suitable for long-term deep learning project maintenance. Regardless of the chosen approach, it's recommended to thoroughly understand version dependencies and maintain environment backups before implementation, ensuring development process stability and reproducibility.