valkyrjaio / valkyrja-starter-app-ts / 30188217879

100%

push

github

[CI] Add a sindri-vitest check that tests the app against sindri-generated classes (#74)

# Description

The application's data-cache classes (`App*RoutingData`,
`AppContainerData`,
`AppEventData`) are produced by Sindri. To compile without running the
generator,
the app commits **empty `*.example.ts` stubs**, and the normal `vitest`
job seeds
those stubs (via the root `postinstall`) and tests against them.
Consequences:

- the normal test suite only ever exercises the empty stubs, never real
generator
  output;
- a generator change — or a framework release that changes what gets
generated —
  is not caught by the app's own tests.

This adds a **second, isolated** CI job that runs Sindri to produce the
real
`App*Data` classes into the source tree and tests the application as it
actually
runs in production. It is kept separate from the stub-based `vitest` job
so
stubbed and generated results never mingle (no fake passes, no mixed
results), and
it doubles as a real-application integration guard for Sindri itself — a
way to
exercise the generator against a real app before cutting a framework
release.

**It cannot reuse the shared `_ts-vitest.yml` workflow:** that workflow
seeds the
stubs via the root `postinstall` and has no step to run the generator,
and its run
step is hard-wired to the `.github/ci/vitest` script. So the job is
defined inline
in this repo's `ci.yml` — **no shared/reusable workflow is modified**,
and the
normal `vitest` (stub) job is left exactly as-is.

## Upgrade to 26.3.0

The Sindri generator fix ([valkyrjaio/sindri-ts#62]) shipped in
`@valkyrjaio/sindri@26.3.0`, so this PR bumps `@valkyrjaio/valkyrja` and
`@valkyrjaio/sindri` to `^26.3.0` and the `sindri-vitest` job runs
green: the
generator now emits populated data caches and the populated assertions
pass. (On
the previous `26.2.0` the generator emitted empty caches — the very bug
#62 fixes
— so this check would have failed, which is exactly what it is meant to
catch.)
The normal ... (continued)

8 of 8 branches covered (100.0%)

Branch coverage included in aggregate %.

72 of 72 relevant lines covered (100.0%)

2.11 hits per line