AlCheck Demo Tour
:ok
{:module, Tour, <<70, 79, 82, 49, 0, 0, 26, ...>>, ...}
A healthy project: `check` runs `format`, `compile`, `compile_test`,
`credo`, `credo_strict`, `dialyzer`, and `test` in parallel — and
finishes by comparing current coverage against the baseline stored on
the `cover_do_not_delete` branch (`baseline_cmd` in `.check.json`).
The footer prints `+X% / -X% / same as baseline` so a regression is
visible at a glance.
`check --only test` runs the test suite alone, partitioned in parallel
per `.check.json`. Useful when you're iterating on test changes.
Now we restore the planted defects and run the fast subset (`format`, `compile`, `credo`). You
see exactly how al_check renders failures: each check fans out, those
that fail flip red, and the failure details follow below.
`check --coverage` runs the coverage calculation, merges partition
results, and prints a per-module summary. Re-running with no source
changes hits the cache instead of re-running calcs.
The `modified_tests` builtin compares the working tree against the base
branch (set via `"base_branch"` in `.check.json`) and runs only the test
lines that changed. `--repeat 10` re-runs those tests up to 10 times via
`mix test --repeat-until-failure`, useful for hunting flakes. Quiet
output here just means nothing has changed relative to `main`; if
nothing's modified, there's nothing to repeat.
`--only test --partitions 5` splits the suite across 5 OS processes
running in parallel. Per-partition logs land in `.check/`.
By default al_check captures test output and prints only summaries.
`--verbose` lets `mix test` write straight to the terminal.