Task execution limits being explicitly tied to compute time

Task timeouts are being made more transparent. By explicitly labeling limits as compute time rather than total duration, developers can better understand how their tasks consume resources. This distinction matters when tasks involve long periods of waiting or sleeping, as wall-clock time no longer creates confusion around timeout rules.

The system is deprecating the generic duration limit in favor of a specific compute seconds parameter. Users can clearly see that idle time spent waiting for external systems does not count against their task execution limits.

Under the hood, the execution engine maps the new compute parameter to the existing duration checks while keeping the old configuration fully supported for backwards compatibility.

Here is how you configure the new compute time limit on a task:

1export const heavyTask = task({
2  id: "heavy-task",
3  maxComputeSeconds: 300,
4  run: async (payload, { ctx }) => {}
5});