3. Run in-process

Status

Current

Context

bluesky code can be run in several ways:

• By the user at an interactive shell, directly calling the run engine in-process.

• Underneath run_plan() if a plan/plan stub is to be run programmatically, which in turn calls the run engine.

• By a central worker process, to which the user would “submit” plans to run.

  • See DLS’s blueapi for an example of a REST API for submitting plans to the run engine.

Present

Tom & Kathryn

Decision

We will run bluesky plans in-process for now, while not excluding the possibility that they could be run behind a worker process at some point in future.

Consequences

• We will not, at least initially, have to write or use an extra worker process.

• Users will have the ability to call the run engine directly

  • This gives us more flexibility initially

  • Makes plotting easier

  • bluesky code will be less isolated from anything else running in-process