Thanks for a wonderful call today. We’ve got a clear way forward on getting to “v1” for workflow. Here it is:
The Functional Requirements will continue to grow, be debated, etc., but they will not impact the launch of v1… most of the hotly debated parts of the functional requirements relate to the features/UX provided by various candidate workflow engines (Camunda, OpenFn, n8n, Tray.io, Joget, etc.) but those specs need not be defined in the API spec and JSON schemas. See point 2.
We must finish the open api spec for 4 endpoints to launch v1:
2.1 list processes in a workflow engine
2.2 describe process, given a process identifier (should return information on what input is required to start that process and what that process does)
2.3 start a process instance (given a process ID and some input data)
2.4 check the status of an instance (or instances)
We must finish the JSON schema for the resources in the API, but this does not mean that we need a full JSON schema for a workflow process, all of its steps, etc., etc. We must define the data that is returned when we “describe” a process, but that data on the process flow may be a text string.
We want to have this done in the next two weeks—it’s already mostly defined in various places in the spec, but needs re-arrangement. I’ll follow up with folks on Slack to assign specific sections. Very excited to drive forward here!