Merge pull request #215 from yuvipanda/upgrade

Upgrade package versions
This commit is contained in:
Yuvi Panda
2018-10-29 13:35:56 -07:00
committed by GitHub
4 changed files with 57 additions and 8 deletions

View File

@@ -0,0 +1,48 @@
.. _contributing/packages:
=======================
Environments & Packages
=======================
TLJH installs packages from different sources during installation.
This document describes the various sources and how to upgrade
versions of packages installed.
Python Environments
===================
TLJH sets up two python environments during installation.
1. **Hub Environment**. JupyterHub, authenticators, spawners, TLJH plugins
and the TLJH configuration management code is installed into this
environment. A `venv <https://docs.python.org/3/library/venv.html>`_ is used,
primarily since conda does not support ARM CPUs and we'd like to support the
RaspberryPI someday. Admins generally do not install custom packages
in this environment.
2. **User Environment**. Jupyter Notebook, JupyterLab, nteract, kernels,
and packages the users wanna use (such as numpy, scipy, etc) are installed
here. A `conda <https://conda.io>`_ environment is used here, since
a lot of scientific packages are available from Conda. ``pip`` is still
used to install Jupyter specific packages, primarily because most notebook
extensions are still available only on `PyPI <https://pypi.org>`_.
Admins can install packages here for use by all users.
Python package versions
=======================
In ``installer.py``, most Python packages have a version specified. This
can be upgraded freely whenever needed. Some of them have version checks
in ``integration-tests/test_extensions.py``, so those might need
updating too.
Apt packages
============
Base operating system packages, including Python itself, are installed
via ``apt`` from the base Ubuntu repositories. The one exception to this
is nodejs, which is installed from the `nodesource <https://github.com/nodesource/distributions>`_
apt repository. The Ubuntu provided version of nodejs is usually too old.
We generally do not pin versions of packages provided by apt, instead
just using the latest versions provided by Ubuntu.

View File

@@ -143,4 +143,5 @@ to people contributing in various ways.
contributing/code-review
contributing/dev-setup
contributing/tests
contributing/plugins
contributing/plugins
contributing/packages

View File

@@ -13,9 +13,9 @@ def test_serverextensions():
], stderr=subprocess.PIPE)
extensions = [
'jupyterlab 0.34.1',
'jupyterlab 0.35.3',
'nbgitpuller 0.6.1',
'nteract_on_jupyter 1.8.1',
'nteract_on_jupyter 1.9.12',
'nbresuse '
]

View File

@@ -185,7 +185,7 @@ def ensure_jupyterhub_package(prefix):
and conda packages!
"""
conda.ensure_pip_packages(prefix, [
'jupyterhub==0.9.2',
'jupyterhub==0.9.4',
'jupyterhub-dummyauthenticator==0.3.1',
'jupyterhub-systemdspawner==0.11',
'jupyterhub-firstuseauthenticator==0.11',
@@ -237,11 +237,11 @@ def ensure_user_environment(user_requirements_txt_file):
conda.ensure_pip_packages(USER_ENV_PREFIX, [
# JupyterHub + notebook package are base requirements for user environment
'jupyterhub==0.9.2',
'notebook==5.6.0',
'jupyterhub==0.9.4',
'notebook==5.7.0',
# Install additional notebook frontends!
'jupyterlab==0.34.1',
'nteract-on-jupyter==1.8.1',
'jupyterlab==0.35.3',
'nteract-on-jupyter==1.9.12',
# nbgitpuller for easily pulling in Git repositories
'nbgitpuller==0.6.1',
# nbresuse to show people how much RAM they are using