b-data / jupyterlab-python-docker-stack

(GPU accelerated) Multi-arch (linux/amd64, linux/arm64/v8) JupyterLab Python docker images. Please submit Pull Requests to the GitLab repository. Mirror of

Home Page:https://gitlab.com/b-data/jupyterlab/python/docker-stack

Repository from Github https://github.comb-data/jupyterlab-python-docker-stackRepository from Github https://github.comb-data/jupyterlab-python-docker-stack

"CONDA_DIR"

bjornjorgensen opened this issue · comments

I don't think this will work

cnf_file = os.path.join(os.getenv("CONDA_DIR", "/usr/lib"), "ssl", "openssl.cnf")

It does.

In Python:

>>> import os
>>> cnf_file = os.path.join(os.getenv("CONDA_DIR", "/usr/lib"), "ssl", "openssl.cnf")
>>> cnf_file
'/usr/lib/ssl/openssl.cnf'

In Terminal

cat /usr/lib/ssl/openssl.cnf

#
# OpenSSL example configuration file.
# See doc/man5/config.pod for more info.
#
# This is mostly being used for generation of certificate requests,
# but may be used for auto loading of providers
...

ok, so it defaults to "/usr/lib" but what about removing "CONDA_DIR" ?

ok, so it defaults to "/usr/lib" but what about removing "CONDA_DIR" ?

Will do so when updating to JupyterLab 4.x. Also including jupyter/docker-stacks#1984.

I have not yet updated to JupyterLab 4.x, because some essential extensions are not yet compatible.
ℹ️ See jupyterlab/jupyterlab#14590 for more information.

ok, what are those essential extensions?

ok, what are those essential extensions?

  • jupyter-server-proxy
  • jupyterlab-git
  • jupyterlab-lsp

See

## Install JupyterLab
RUN export PIP_BREAK_SYSTEM_PACKAGES=1 \
&& pip install \
jupyter-server-proxy \
jupyterhub==${JUPYTERHUB_VERSION} \
jupyterlab==${JUPYTERLAB_VERSION} \
jupyterlab-git \
jupyterlab-lsp \
notebook \
nbconvert \
python-lsp-server[all] \

ok, thank you.
Well, my idea it to use debian testing or unstable, have it with python 3.12 RC (for testing) and java 21. Remove coder-server and only have a basic jupyterlab and not having GPU stuff. I use Visual Studio Code Insiders as IDE and the docker files are on a server. So I connect to it with remote ssh.

Well, my idea it to use debian testing or unstable, have it with python 3.12 RC (for testing) and java 21. Remove coder-server and only have a basic jupyterlab and not having GPU stuff. I use Visual Studio Code Insiders as IDE and the docker files are on a server. So I connect to it with remote ssh.

Then, it is better you build on https://github.com/b-data/python-docker-stack.

Be aware of https://github.com/b-data/python-docker-stack/blob/main/ver/latest.Dockerfile#L9 and that there is neither
glcr.b-data.ch/python/psi/3.12.0rc<n>/debian:testing nor
glcr.b-data.ch/python/psi/3.12.0rc<n>/debian:unstable

Then, it is better you build on https://github.com/b-data/python-docker-stack.

That is what I do for the Data Science Dev Containers: https://github.com/b-data/data-science-devcontainers

Will do so when updating to JupyterLab 4.x. Also including jupyter/docker-stacks#1984.

Not including jupyter/docker-stacks#2006 and jupyter/docker-stacks#2007, though.
👉 I.e. not going to migrate startup scripts and hooks to Python.

Closed by 536a918