I want to be able to host my own open data platform so others can use the code and datasets my team creates

[petebachant]

It seems like some teams are reinventing the wheel here, creating bespoke websites to share their stuff. If they don't want to post it on a common server, we can at least make it easy for folks to host their own, and the goal of this software is to make the data easily usable.

In other words, open source this package so others can self-host, but enable some sort of federated search so projects and artifacts can be found on any instance. We also want commands like calkit import dataset ... to work with other domains.

[petebachant]

Notes from WindLab webinar:

• Trying to make it easy for people to validate their models with predefined cases. Can we create Calkit projects that people use as a template for this sort of benchmarking, where they automatically import the dataset and some pipeline stage that compares the results. Where the users need to create an intermediate stage that saves their data in the correct format?
  • Maybe a good use case for a CAD feature? At least the CAD models should be part of the repo, hosted on Figshare or whatever.
• Challenge of separating commercial companies' controller software, which needs to be closed source. Uses HTTP server. They use the open source model as a digital twin so commercial companies can test their designs. Use case for making some files private, or perhaps referencing a model in a private project.

WindLab

• What is WindLab?
• https://windlab.hlrs.de/
• "Harvesting" plugin to collect metadata from related CKAN instances? What is CKAN? https://github.com/ckan/ckan
• Very interesting. An open source data management system.
• WindLab is a CKAN instance. So is AIRE.
• We should probably be interoperable with CKAN instances, letting them "harvest" datasets from Calkit instances.
• We could also list CKAN instances in Calkit.
• We should allow users to import datasets from CKAN instances and cache them with DVC. --> ckanfs?
• Goal is not to duplicate data
• WindIO schema? What is this?
• Python package for WindLab will go into GitLab, but where?

FLOW

• "Model chain" -- is this a pipeline?
• FLOW seems to be like Calkit
• Download Jupyter notebooks
• Has linking between processes, publications, datasets

AIRE

• 8 experimental wind farms, 4 are commercial.
• Blade erosion studies. Measuring precipitation.
• Data and knowledge, models, tools, case studies, component designs --> Calkit models for all these?