asyncExecution property
Runs the test asynchronously, by enqueueing it through the real execution
queue under trigger_id instead of running the task graph inline.
Default false, which preserves the existing synchronous behavior. The
enqueued request carries no caller identity, so it is drained under the
integration's run-as service account, exactly as a real Cloud Scheduler,
cron, Pub/Sub, Salesforce or connector fire would be. That is the only way
to verify the credential path an asynchronous trigger actually takes; a
synchronous test always carries the caller's own credential and so
exercises the opposite branch. When true: * The version named by
integration_version.name runs. A name ending in /versions/{version}
runs that exact version, published or draft; a name without one runs the
published (ACTIVE) version on the trigger. Either way the integration is
resolved from storage, so integration_version must carry only name;
supplying any other field is rejected, rather than silently running
something other than what was passed. * A draft is validated before it is
enqueued, including the check that the caller may act as the integration's
run-as service account. Testing a draft therefore fails fast on a
configuration a publish would reject. * The response carries
execution_id only. execution_failed, event_parameters, parameters
and parameter_entries describe a finished run and are left unset,
because nothing has run yet. In particular, do not read execution_failed
as a pass signal. * The run is real, with real side effects, quota and
concurrency. No actual Pub/Sub push or Cloud Scheduler tick occurs; only
the enqueue is simulated, which does not change the credential path.
Optional.
Implementation
core.bool? asyncExecution;