Comment by Cthulhu_

Yeah I think what you're describing is an issue for all "traditional" database migrations, that is, switching between branches is an issue.

I think the best thing to do is, during development, recreate and re-populate the database on every branch change automatically. I have that set up in my current project, although I have to remove the database file manually for now. My test data is hardcoded into my codebase, using the repository access code to save the data on startup, if a "load test data" flag is set.

of course, using an in-memory database would also resolve it, but then any changes I make in the data at runtime would be lost every time.