Courts want confidence from CMS extract through gold calendar publish before promoting a change. Where should end-to-end tests run?

Courts that want confidence from CMS extract through gold calendar publish before promoting a change should run end-to-end tests in a non-production workspace along that full path. E2E tests validate the assembled pipeline—extract, transforms, dependencies, and gold publish—under realistic orchestration, catching interface mismatches that unit tests miss. Justice calendars are public-facing; promoting untested extract-to-gold paths risks wrong hearing data.

Running end-to-end tests from CMS extract through silver layer validation in a non-production workspace fails because it stops at silver validation and never proves the full extract-to-gold publish path. Running end-to-end tests from mock CMS extracts through gold calendar publish in a non-production workspace fails because mocked extracts don't prove the real CMS extract connection and miss extract-level failures. Running end-to-end tests from CMS extract through gold calendar publish in a production workspace sandbox fails because even a sandbox in production still puts unvalidated orchestration beside live dockets instead of proving the path in non-prod first.

Exam caveat: E2E belongs in non-prod (or carefully isolated test projects), after unit/integration successes, before prod promote. E2E is broader than integration: it stresses the full orchestrated path and often the job DAG itself. UAT may still follow for clerk-facing acceptance.

Operational check: in a non-prod workspace, run E2E from CMS extract through gold calendar publish, assert key gold invariants (counts, date ranges, courtroom attributes), and only then promote. Confirm production dockets are not the primary E2E laboratory.