Many applications are automatically tested on each commit inside a GitLab pipeline. If your application relies on a database such as PostgreSQL, it can be tempting to use an in-memory database such as H2 for tests because it is easier to set up and destroy for each execution of your test suite. While this works for simple applications, it is no longer possible once you rely on vendor-specific SQL features (e.g., usage of column types not available in H2). In ...
Traditionally, relational databases required all tables to have a fixed data schema, i.e. a set of attributes (such as a “user” table with attributes “firstName”, “lastName” and “email”). You could change the schema by adding and removing attributes, but the available attributes were always know at a given point in time. For many use cases, this was acceptable. However, some applications have the need to store semi-structured data: entries where the attributes are not known ahead of time. Since 2012, ...