Some aspects cannot be determined just by reading code. Whether a form submits, whether a workflow completes, or whether a page renders — these require the application to be running and actively used.
Browser tests enable Coroid to verify these aspects. You can find them under Quality → Tests → Browser.
Scenarios
A scenario outlines a user workflow in plain language: what actions to take and what should be true afterwards.
Sign in as a standard user, open the billing page, and change the
plan from Free to Professional.
Expect: the confirmation dialog lists the new monthly price, and after
confirming, the plan shown in the header is Professional.Define expectations as observable outcomes. Saying "the page works" offers no meaningful way to detect failure; stating "the header displays Professional" does.
Test targets and accounts
A test target is the environment where a scenario executes — typically a preview of the change being tested.
Each target has its own configuration and test accounts. Test accounts are credentials created specifically for this purpose.
Gating on results
Browser test results can feed a quality gate just like any other check — either as a blocker or an advisory measure.
Advisory is the appropriate default setting here, more so than elsewhere. Browser tests are the most unreliable type of check in any system: they rely on timing, network conditions, and a live environment. Only make them blocking once a specific scenario proves stable over multiple runs.
Keeping them useful
- Few and important. Ten scenarios covering critical workflows are better than a hundred covering everything. Each scenario adds time to every execution.
- Independent. A scenario that relies on another running first will produce confusing failures.
- Clean up. A scenario that creates data must not leave behind state that disrupts the next run.
When a scenario fails
Examine the evidence before looking at the code. Browser runs record what occurred, and the most common causes are environmental — the preview wasn’t ready, a dependency was down, or an account was locked.
A genuine failure will occur repeatedly. A flaky result will not.