Azure Ubuntu 16.04虚拟机中升级Python3.5至Python3.6的方法及JupyterLab可行性咨询

Hey there! Let's fix that "unable to locate package" error and get Python 3.6 set up on your VM, plus cover how to handle this with JupyterLab:

Why apt-get can't find Python 3.6? #

Ubuntu 16.04's default software repositories don't include Python 3.6 out of the box. You'll need to add a trusted third-party PPA (Personal Package Archive) that hosts older/newer Python versions—deadsnakes is the standard choice for this.

Method 1: System-wide install via APT #

Follow these steps to get Python 3.6 working with apt:

1. Add the deadsnakes PPA to your system:

   sudo add-apt-repository ppa:deadsnakes/ppa
   

   Press Enter when prompted to confirm adding the repository.
2. Update your package lists to pull in the new PPA's contents:

   sudo apt-get update
   

3. Install Python 3.6:

   sudo apt-get install python3.6
   

   This should resolve the package not found error now.

Optional: Set Python 3.6 as default python3 #

If you want python3 to point to 3.6 instead of the original 3.5, use the update-alternatives tool:

# Register both versions with priority levels
sudo update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.5 1
sudo update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.6 2

# Select which version to use by default
sudo update-alternatives --config python3

Just enter the number corresponding to Python 3.6 when prompted, and you're done.

Method 2: Install via Anaconda (Recommended for your setup) #

Since your current Python is from Anaconda, creating a dedicated environment with Python 3.6 is safer (it won't mess with your system's default Python):

1. Create a new conda environment with Python 3.6:

   conda create -n py36 python=3.6
   

2. Activate the environment to start using Python 3.6:

   source activate py36
   

   Now running python --version will show Python 3.6 in this environment.

Can you do this via JupyterLab? #

Absolutely! Here's how to add your Python 3.6 environment as a Jupyter kernel:

For conda environments: #

1. Activate your py36 environment first:

   source activate py36
   

2. Install the ipykernel package to connect the environment to Jupyter:

   conda install ipykernel
   

3. Register the environment as a Jupyter kernel:

   python -m ipykernel install --user --name py36 --display-name "Python 3.6 (py36)"
   

4. Restart JupyterLab, and you'll see the "Python 3.6 (py36)" option in the kernel launcher.

For APT-installed Python 3.6: #

1. Install required tools for kernel setup:

   sudo apt-get install python3.6-dev python3.6-pip
   pip3.6 install ipykernel
   

2. Register the kernel:

   python3.6 -m ipykernel install --user --name python36 --display-name "Python 3.6"
   

   Restart JupyterLab, and the new kernel will be available.

内容的提问来源于stack exchange，提问作者blackraven