clawdie-ai/setup/preflight.test.md

preflight.test.ts — test suite summary

Smoke tests for the preflight orchestrator (setup/preflight.ts), which runs the 12 setup steps before installation.

Setup / mocks

• spawnSync is stubbed so no real subprocesses run — every call returns exit 0 by default
• fs.writeFileSync / mkdirSync are stubbed to capture what gets written without touching disk
• process.exit is spied on so a failing preflight does not kill the test runner
• Two helpers extract the outputs: getSummaryJson() and getSummaryEnv() parse what was written to summary.json / summary.env

5 describe blocks, 17 tests

basic run (no flags)

All 12 steps run in the correct order:

environment → pi-config → jails → db → git → cms →
hosts → mounts → telegram-auth → service → verify → doctor

• summary.json has overallStatus: success
• summary.env has OVERALL_STATUS=success, WITH_ONBOARDING=false, CAPTURE_PASSWORD_STEP=false
• process.exit is not called

failure handling

• A single failing step does not abort the run — all 12 still execute
• overallStatus becomes failed
• process.exit(1) is called
• --fail-fast stops after the first failure (only 1 npm call, then exit 1)

--with-onboarding

• The interactive onboarding step skips gracefully in a non-TTY environment (test runner has no TTY)
• All 12 non-interactive steps still run
• overallStatus is failed because onboarding itself fails without a TTY
• WITH_ONBOARDING=true appears in summary.env

--capture-password-step

• Implies --with-onboarding (verified via summary.json args)
• The Python screenshot capture does NOT run when onboarding failed (no TTY)
• CAPTURE_PASSWORD_STEP=true appears in summary.env

summary output structure

• Every step result in summary.json has all required fields: id, label, commandLine, exitCode, status, startedAt, finishedAt, logFile
• Log file paths are bare filenames with no path separator (e.g. environment.log)
• summary.env contains a *_STATUS= and *_EXIT_CODE= line for every step