Reporters¶
FastProof accepts the report files your tests already write (see Report formats). Its own reporters go further: they send results while tests run, stream logs live, keep every retry as an attempt, and never fail your test run when the server is unreachable.
| Reporter | Package | Status |
|---|---|---|
| Playwright | @franshub/fastproof-playwright (npm) |
First release in preparation |
| pytest | pytest-fastproof (PyPI) |
First release in preparation |
More frameworks follow; until then, upload the files of the framework's own reporter with fastproof upload.
What they do¶
- Live results: a test shows up as
runningwhen it starts and gets its status when it ends. - Live logs: stdout, stderr, logging and steps reach the launch page while the test runs (see Live logs).
- Attempts: each retry is stored as an attempt of the same result, with its own log and attachments.
- Attachments: screenshots, videos, traces and attached files are uploaded with the result.
- Trace correlation: each attempt carries a W3C
traceparent, so calls to FastMock and traced services can be matched to the test that made them. - Secret masking at the source: values of environment variables that look like secrets (
*TOKEN*,*SECRET*,*PASSWORD*,*_KEYand similar) and your own patterns are masked before anything is sent. - Offline spool: when the server is down or slow, everything goes to a bounded spool on disk (default
.fastproof/spool, 200 MB) and is retried with backoff; the next run sends what is left, or you send it yourself. A reporter problem never changes the result of the test run. - CI aware: the launch name and the external launch id come from the CI system, so parallel shards and jobs of one pipeline join one launch.
- Test plans: an Allure-format
testplan.json(FASTPROOF_TESTPLAN) selects which tests run.
Common settings¶
Both reporters read the same environment variables as the fastproof CLI; without FASTPROOF_URL they stay
off.
| Variable | Meaning |
|---|---|
FASTPROOF_URL |
Server URL |
FASTPROOF_TOKEN, FASTPROOF_TOKEN_FILE |
API token with launches:write (or a file holding it) |
FASTPROOF_PROJECT |
Project key |
FASTPROOF_LAUNCH |
External launch id; runs with the same id join one launch (default: from CI) |
FASTPROOF_LAUNCH_NAME |
Launch name |
FASTPROOF_ENVIRONMENT |
Environment key, part of the test identity |
FASTPROOF_ENABLED |
false turns the reporter off |
FASTPROOF_OFFLINE |
true writes only the spool and sends nothing |
FASTPROOF_SPOOL_DIR, FASTPROOF_SPOOL_MAX_BYTES |
Spool location and size limit |
FASTPROOF_FLUSH_TIMEOUT_MS |
How long the end of the run waits for the last uploads (default 30 s) |
FASTPROOF_LIVE_LOGS |
false sends logs with the result instead of live |
FASTPROOF_SECRET_ENV, FASTPROOF_SECRET_PATTERNS |
Extra environment variable names (globs) and regular expressions to mask |
FASTPROOF_TESTPLAN |
Path of an Allure-format testplan.json |
FASTPROOF_QUIET |
true prints nothing but errors |
Give the token through your CI system's secret store, never in the repository.
Playwright¶
// playwright.config.ts
import { defineConfig } from '@playwright/test';
export default defineConfig({
retries: 2,
reporter: [
['list'],
['@franshub/fastproof-playwright', { environment: 'staging' }],
],
});
Options in the config override the environment. Playwright 1.45 or newer and Node.js 20 or newer are
supported. Keep your usual reporters: FastProof's does not print the test list. The Playwright retry number
becomes the attempt, and test.step steps and failed expect calls are streamed as steps. The
@franshub/fastproof-playwright/fixture export adds a test with trace headers for outgoing requests.
Send what earlier runs left in the spool:
pytest¶
FASTPROOF_URL=https://fastproof.example.com FASTPROOF_PROJECT=shop \
pytest --fastproof-environment staging
The plugin registers itself; settings are also --fastproof-* options (--fastproof-url,
--fastproof-project, --fastproof-launch, --fastproof-offline, --fastproof-disable and others; see
pytest --help). pytest-xdist works: workers stream logs, and the controller sends results once. Live
streaming covers logging records and steps; captured stdout and stderr arrive per test phase (or live with
-s). Lower the log level (log_level = INFO in the ini file) to stream more than warnings.
Send what earlier runs left in the spool: