Attain competency in `DataPackages`

[bkuczenski]

DataPackages from frictionless data are the preferred format for sharing structured data generated in the hackathon. One role of the reproducibility group will be to ensure that the work products of the other groups can be easily packaged in this manner for distribution to other users.

The first task in this objective is to become familiar with how datapackages work, determine whether we need to create a custom schema or use the existing ones, and understand clearly how to package the work output from another group into this format.

This issue will be closed when the first output object from another working group is rendered as a DataPackage using an automated workflow.

[mfastudillo]

I had a very quick go to the basics some days ago, based on this example. But there is basic things that I don't know yet how to do.

[massimopizzol]

Alright I worked on this and created the feature-datapack-cortabs branch of the correspondence table repo, I also filed a pull request.

• prepared an example tutorial on how to generate metadata with datapackage. Check it out. A key issue is that the type of data (string, float, number) must be checked carefully after using the 'infer' function. The program didn't get it right in this case.

• My understanding is that we can either make one datapackage.json file describing all corresponding tables at once, or we can make one datapackage.json file for each correspondance table. In the latter case we should create a specific directory as I did here as an example. This seems to me a good way to go.

• I have proposed a list of properties for the descriptor, based on the recommended mandatory and optional ones.

[bkuczenski]

Massimo, this is great! Can you talk about this at the 9am check-in for a few minutes?