Emits one console.* call or uncaught error from the page under test. Rendered verbatim — no # prefix — because it is the page's output, not qunitx's, and prefixing it would corrupt whatever the page was deliberately printing.

Reporter wiring. config.reporter selects exactly one stdout reporter; artifact reporters (JUnit) are additive and stack on top. Built once per run in Config.setup and shared by every concurrent group — the group configs are spread off the parent config, so they all reference this same array (the same way the run counter is shared).

A diagnostic that also belongs on stderr — a stack trace, a timeout, a page that crashed.

A decision the runner made and wants on the record — what it chose to run, what it skipped.

Emits run end to every active reporter, awaiting any that flush asynchronously.

Emits run start to every active reporter. In watch mode this fires once per rerun.

Applies one testEnd to the counters, then fans it out to every active reporter. The counter update happens here — exactly once, before any reporter runs — so counts stay correct regardless of how many reporters are attached.

Something surprising that did not stop the run — a filter that matched nothing, a flag that does not apply to the chosen browser.

Options a diagnostic can carry beyond its text. Both default to the CLI's long-standing behaviour, so only the handful of call sites that need something else say so.

The --reporter value -> its class. Keyed by ReporterName, so adding a name to REPORTERS without wiring it up here is a type error rather than a silent fall back to tap.