Coordinating parallelizable work from a task?

[opsound]

I read the Hubris reference and am having trouble envisioning how to achieve certain behavior given the synchronous IPC constraint:

Sending a message always blocks the sender, waiting for a reply

Suppose task A (orchestrator) needs to perform work which could be run in parallel by tasks B and C (say they each abstract access to some separate HW).

Is there a way to achieve this? Or an alternative task responsibility pattern to make it possible?

EDIT - reading further, it seems like notifications pingback pattern could work.

The task A orchestrator ("client" here) can send to multiple servers (tasks B and C) to kick off the work, and await notifications from each, which represent the completion of said work, and then send again to B and C retrieve the result of the work?

[steveklabnik]

Yep, your edit is how I would structure this. The POST syscall would let B and C send those notifications to A.