All notable changes to this project will be documented in this file.
- Remove Python 3.8 support
- Added Python 3.11 support
- Switch all images to CentOS Stream 9 (c9s)
- Use new base image https://quay.io/repository/sclorg/python-39-c9s
- Runtimes set now before workbench images to simplify chaining
- Toolkit installed in CUDA Development images
- Add PyTorch and Tensorflow images because of the new c9s base
- Fix #28
- Updates to apps and packages
- VSCode: 4.9.1 => 4.16.1
- R: 4.2.2 => 4.3.1
- RStudio: 2022.12.0-353 => 2023.06.1-524
- Fixes on VSCode and RStudio images for bad redirects, integrated in 2023b
- Addition of TrustyAI workbench image, integrated in 2023b
- Addition of runtime images, integrated in 2023b
- 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.
- 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.
- 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.
- Minimal:
- 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
- Minimal:
- 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
- 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.
- 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 innotebook-controller-images
that works only with the Kubeflow Notebook Controller and is not compatible with ODH/RHODS dashboard).
- Monai notebook in the
notebook-controller-images
line. As it's not thoroughly tested it's still not officially released - OptaPy custom image updated with OptaPy 8.23.0a0, minimal-data-science-optapy-notebook:v0.0.2
- All images, including those in the
jupyterhub-images
folder are now compatible with the Kubeflow Notebook Controller
- 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
- New notebook image: CUDA 11.6.2 + Monai 0.9.0 + Streamlit 1.10.0 notebook (quay.io/guimou/monai-notebook:0.0.2)
- 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/generic-data-science-streamlit-notebook:0.0.7
- nbgitpuller added to Streamlit notebook (quay.io/guimou/generic-data-science-streamlit-notebook:0.0.6)
- fix yarn cache clean on Streamlit notebook image
- prebuilt-images repos added to CHANGELOG
- Streamlit notebook (quay.io/guimou/generic-data-science-streamlit-notebook:0.0.5)
- Changelog file
- SageMath notebook (quay.io/guimou/minimal-data-science-sagemath-notebook:0.0.1)
- RStudio added to the R notebook (quay.io/guimou/odh-minimal-data-science-r-notebook:0.0.15)
- OptaPy notebook added (quay.io/guimou/minimal-data-science-optapy-notebook:v0.0.1)