Currently, changes on the DB schema will only work on fresh installs of the project.
This can be easily tested by:
1) Running extract_features + generate_matches (either through the UI or scripts) on Dataset A
2) Changing the schema of a given table (in my specific case it was adding a new field to the matches table, but it could easy to reproduce by adding a non-nullable field to any table)
3) Running extract_features + generate_matches (either through the UI or scripts) on Dataset B
Currently, changes on the DB schema will only work on fresh installs of the project.
This can be easily tested by:
1) Running extract_features + generate_matches (either through the UI or scripts) on Dataset A 2) Changing the schema of a given table (in my specific case it was adding a new field to the matches table, but it could easy to reproduce by adding a non-nullable field to any table) 3) Running extract_features + generate_matches (either through the UI or scripts) on Dataset B
John's suggestion could be an interesting one (https://github.com/miguelgrinberg/Flask-Migrate).
I investigated using it, however, I ran into some issues