2023c

Main changes for this release are:

• All images are now based on CentOS Stream 9 to allow access to a broader range of OS packages than what UBI allows
• Base images (standard and CUDA) include additional libraries and applications that were often requested (libsndfile, ffmpeg, gstreamer, various codecs,...)
• The base images also include standard Databases clients: MySQL, PostgreSQL, MSSQL, MongoDB,...
• For almost all images, Python 3.11 (now the default) and Python 3.9 versions are available. Python 3.8 support is dropped.
• All packages in the various flavours have been updated to their (almost) latest versions.
• A new flavour is introduced, LangChain, for you LLMs lovers! Everything is bundled and ready to use.
• On top of latest VSCode and RStudio version, a new IDE is introduced, Langflow, to allow quick prototyping LangChain pipelines in UI-mode.
• Images are now more modular, allowing you to build your own new custom image. Choose you Base (Standard, CUDA,...), your Protein (aka Bundle from Data Science, PyTorch, Tensorflow, LangChain, Spark, R,...), and your favourite Topping (aka IDE from Jupyter, VSCode, RStudio, Langflow) and let the magic happen.
• Why magic? Well, because there is now a Wizard to guide you through the process of creating those custom images (CLI-based for now, Web UI coming soon).
• With this new modular approach, it will also be much easier for anyone to create (and share!) their own recipes for the Bundles or IDEs they love.

Release 2023b

Most important update

Update to RStudio and VSCode images (standalone versions). They are now fully compatible with the dashboard, so can be used as standard custom images. Idle culling also works with those images.

Other updates

• Update to the CUDA RStudio and Jupyter + RStudio images to include the CUDA toolkit. This is necessary to install certain packages that are compiled in-place.
• Addition of some packages:
  • Minimal:
    • jupyter-server-terminals ~=0.4.4 and jupyter-server ~=2.1.0: this will allow culling to work for terminals once enable on the notebook controller.
• Updates of some packages versions:
  • Minimal:
    • jupyterlab ~=3.5.2 -> ~=3.5.3
    • jupyter-resource-usage: ~=0.6.4 -> ~=0.7.0
  • Datascience:
    • matplotlib ~=3.6.2 -> ~=3.6.3
    • pandas ~=1.5.2 -> ~=1.5.3
    • plotly ~=5.11.0 -> ~=5.13.0
    • scikit-learn ~=1.2.0 -> ~=1.2.1
    • scipy ~=1.9.3 -> ~=1.10.0
    • streamlit ~=1.16.0 -> ~=1.17.0

2023a

Major Release, including:

• Migration of repository to opendatahub-io-contrib organization
• Modification of versioning scheme to allow multiple parallel releases and automated regular updates as described here
• Full refactoring of the repository for clarity
• Adding tests to the Makefiles to validate images
• Release of 2023a, creation of associated support branch

0.0.15 - Add Spark notebook

• Spark notebook in the notebook-controller-images folder. This is a JupyterLab Data Science notebook with Spark 3.3.1 and Hadoop 3.3.4.
• Also includes RBAC and examples for launching Spark workloads directly from the notebook.

0.0.14 - Add VSCode to Jupyter images

Added

• VSCode notebook in the jupyterhub-images folder. This is a JupyterLab Data Science notebook that also includes VSCode 4.7.1. This image works both with JupyterHub and current ODH/RHODS dashboard version (contrary to the image in notebook-controller-images that works only with the Kubeflow Notebook Controller and is not compatible with ODH/RHODS dashboard).

0.0.13 - Images update for KFNBC compatibility + Monai image

Added

• Monai notebook in the notebook-controller-images line.

Modified

• All images are now compatible with the Kubeflow Notebook Controller. The images in the jupyterhub-images folder stay compatible with JupyterHub.
• OptaPy custom image updated to OptaPy 8.23.0a0: s2i-minimal-data-science-optapy-notebook:v0.1.0

0.0.12 - KF Notebook Controller images

Added

• New organization of images for upcoming Kubeflow Notebook Controller
• Base, Minimal, Data Science, Code-Server and RStudio images available in the new notebook-controller-images line

0.0.11 - Monai + Streamlit update

Added

• New notebook image: CUDA 11.6.2 + Monai 0.9.0 + Streamlit 1.10.0 notebook (quay.io/guimou/s2i-monai-notebook:0.0.2)

Modified

• Streamlit image updated to streamlit 1.10.0 - fixes #5
• fix Streamlit launch scripts for directory context - fixes #6
• new streamlit image: quay.io/guimou/s2i-generic-data-science-streamlit-notebook:0.0.7

0.0.10 - Add nbgitpuller to Streamlit

Merge pull request #7 from guimou/feature

streamlit: nbgitpuller + fixes

0.0.9 - Addition of Streamlit

Merge pull request #4 from guimou/streamlit

Addition of Streamlit notebook