NBEnvKernels/Env-specific Notebook Kernel.ipynb

Env-specific Notebook Kernels.ipynb

{
  "cells": [
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "# Env-specific Notebook Kernels"
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "**author**: github.com/ruxi\n<br>**created**: 2016-03-06 19:36:28 "
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "## Problem"
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": " - anaconda environments prevent namespace pollution\n - packages installed in ENVs are accessed by kernels\n - jupyter notebook do not install ENV-specific python kernels, but reuse common paths\n - A simple [solution](https://github.com/jupyter/notebook/issues/541) is to create a new kernelspec that points to $PYTHON_PATH\n  "
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "## Intention\n\n**Before:**\n```python\nimport sys; sys.executable\n```\n> '/usr/bin/python'\n\n**After:**\n\n```python\nimport sys; sys.executable\n```\n> ~/anaconda3/<**your env name**>/bin/python"
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "# Method\n\n- create new kernel\n- drop in '/usr/local/share/jupyter/kernels/'\n- F5!"
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "## Create new kernel"
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "**Anatomy of existing kernelspec**\n\n```python\n!cat ~/.local/share/jupyter/kernels/python3/kernel.json\n```\n\n```bash\n{\n \"display_name\": \"Python 3 (anaconda3/bin/python)\",\n \"argv\": [\n  \"~/anaconda3/bin/python\",\n  \"-m\",\n  \"ipykernel\",\n  \"-f\",\n  \"{connection_file}\"\n ],\n \"language\": \"python\"\n}\n```\nReplace <font color='blue'>\"~/anaconda3/bin/python\"</font> with <font color='blue'>python</font>\n"
    },
    {
      "metadata": {
        "collapsed": true,
        "trusted": false
      },
      "cell_type": "code",
      "source": "import os.path\nos.makedirs('env_python_kernel')",
      "execution_count": null,
      "outputs": []
    },
    {
      "metadata": {
        "collapsed": false,
        "trusted": false
      },
      "cell_type": "code",
      "source": "%%writefile env_python_kernel/kernel.json\n{\n \"display_name\": \"PythonPath (python)\",\n \"argv\": [\n  \"python\",\n  \"-m\",\n  \"ipykernel\",\n  \"-f\",\n  \"{connection_file}\"\n ],\n \"language\": \"python\"\n}\n",
      "execution_count": null,
      "outputs": []
    },
    {
      "metadata": {
        "collapsed": false
      },
      "cell_type": "markdown",
      "source": "## List currently kernel list"
    },
    {
      "metadata": {
        "collapsed": false,
        "trusted": true,
        "ExecuteTime": {
          "start_time": "2016-10-07T15:35:46.037502",
          "end_time": "2016-10-07T15:35:47.219658"
        }
      },
      "cell_type": "code",
      "source": " !jupyter-kernelspec list",
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": "Available kernels:\r\n  ruby                /home/lewis/.ipython/kernels/ruby\r\n  ir                  /home/lewis/anaconda3/envs/phd/share/jupyter/kernels/ir\r\n  python2             /usr/local/share/jupyter/kernels/python2\r\n  python3             /usr/local/share/jupyter/kernels/python3\r\n  pythonpathkernel    /usr/local/share/jupyter/kernels/pythonpathkernel\r\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "collapsed": true
      },
      "cell_type": "markdown",
      "source": "## Copy/Paste/Replace kernel.json + F5"
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "## Check if it worked"
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "```\nsource activate <envname>\njupyter notebook\n```"
    },
    {
      "metadata": {
        "collapsed": false,
        "trusted": true,
        "ExecuteTime": {
          "start_time": "2016-10-07T15:30:47.772707",
          "end_time": "2016-10-07T15:30:47.801360"
        }
      },
      "cell_type": "code",
      "source": "import sys; sys.executable #did it work?",
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": "'/home/lewis/anaconda3/envs/phd/bin/python'"
          },
          "metadata": {},
          "execution_count": 2
        }
      ]
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "---\n\n\n# To install new kernels to global\n\n**Python 2.7 example**\n\n```bash\nconda create env -n py27 python==2.7\nsource activate env # for windows: activate py27\njupyter-kernelspec install-self # adds current kernel location to jupyter configs (available to global)\n```\n\n**Note:** \n\n```jupyter kernelspec install-self``` was **DEPRECATED as of jupyter 4.0**. But... still worked\n\nHaven't tried, but error prompt suggests proper method is to use \n```bash\nipython kernel install```"
    }
  ],
  "metadata": {
    "kernelspec": {
      "name": "pythonpathkernel",
      "display_name": "pythonpath (python)",
      "language": "python"
    },
    "toc": {
      "threshold": 6,
      "number_sections": true,
      "toc_cell": false,
      "toc_window_display": true,
      "toc_section_display": "block",
      "sideBar": true,
      "navigate_menu": true
    },
    "_draft": {
      "nbviewer_url": "https://gist.github.com/53c4b8f58dab0a020415d617582d389e"
    },
    "language_info": {
      "nbconvert_exporter": "python",
      "codemirror_mode": {
        "version": 3,
        "name": "ipython"
      },
      "file_extension": ".py",
      "version": "3.5.2",
      "mimetype": "text/x-python",
      "pygments_lexer": "ipython3",
      "name": "python"
    },
    "gist": {
      "id": "53c4b8f58dab0a020415d617582d389e",
      "data": {
        "description": "NBEnvKernels/Env-specific Notebook Kernel.ipynb",
        "public": false
      }
    },
    "latex_envs": {
      "eqNumInitial": 0,
      "eqLabelWithNumbers": true,
      "current_citInitial": 1,
      "cite_by": "apalike",
      "bibliofile": "biblio.bib"
    },
    "nav_menu": {}
  },
  "nbformat": 4,
  "nbformat_minor": 0
}

jupyter --paths

http://stackoverflow.com/questions/37085665/in-which-conda-environment-is-jupyter-executing

conda install -c conda-forge nb_conda_kernels