Automate deployment

[chekos]

datasette doesn't allow you to "update" a deployment. Each time one publishes a new version of a datasette we are effectively overwriting it. This means the database must be included each time.

This is a problem because as of right now, the original dataset is ~650 MB and the "broken down" database is ~1.1 GB (#1 might bring this down).

We want to automate the deployment using GitHub and GitHub Actions to facilitate the development of the analysis of the data set. In the metadata we can include views and queries found to be useful.

Updating these views, queries, and other elements of the metadata should be easy for anyone who wants to be involved in the project.

Challenges:

• The size of the database itself. Even if we hosted on something like an s3 bucket, what are the limits of GitHub Actions?
• How often will we really update the metadata? Could we aggregate updates and deploy at the end of the week instead of instantly?

[chekos]

DONE ✅ .github/workflows/main.yaml puts together original dataset, transforms the data, builds sqlite3 database and deploys to heroku 🚀