Backend development

The Spring Boot application lives in mtl-server/. Maven owns compilation, packaging, and tests. Run these commands inside the existing development container, from mtl-server/:

mvn test
mvn package
mvn spring-boot:run -Dspring-boot.run.profiles=dev

The diagnostics module compiles during the Maven lifecycle and is required at runtime. Use the configured module path; see GPS diagnostics.

Use an isolated database and public/synthetic data for tests. Follow the testing guide and preserve pre-existing failures separately from results for the change.

After API changes, export the live schema and regenerate the client through the API workflow. Source and runtime settings are owned by the POM and application configuration.

Implementation

Keep advanced domain calculations on the server. Change APIs rather than duplicating calculation logic in the client. Prefer existing repositories and standard framework features over separate persistence helper classes.

Format code and organize Java imports using the project style. Limit cleanup to code changed for the task. Follow shared code rules and software version policy.

Logging

Log major server operations and lifecycle phases at INFO: start, meaningful milestones or phase summaries, completion, duration, and useful counts. Keep per-record and other high-volume detail at DEBUG.

Log every caught exception with operation context, its type, and a safe message before recovering, falling back, or returning. Do not silently ignore exceptions or use empty catch blocks. Use WARN for recoverable failures and ERROR with the throwable when the operation fails. Keep repeated recoverable stack traces at DEBUG to avoid flooding normal logs.

Data migration

Do not add legacy-data backfills or old-data migration paths unless explicitly requested. Assume users can recreate the database after derived-statistic or data-shape changes. Keep ingest and recalculation for new imports; avoid migration-only administration endpoints and services.