Checkpoint: Data Layer

This checkpoint grades data-layer reasoning in runnable form: relationship modeling, SQL semantics, transaction judgment, and full-stack integration sense — the decisions a senior reviewer makes on every data-touching PR.

Data outlives code

Frameworks churn; schemas persist for a decade. The checkpoint's exercises — normalizing a table, choosing the join, writing the constraint that makes bad data impossible — are the durable half of full-stack work. Get them right and the application layer almost writes itself; get them wrong and every future feature inherits the mistake.

Think in sets, ship in constraints

Two habits separate SQL fluency from SQL survival: thinking in sets rather than loops, and pushing correctness into the schema so a bad row is impossible rather than merely discouraged. The checkpoint's exercises test both — the join that answers a question in one query, and the foreign key plus CHECK that make the data trustworthy no matter which client writes it.