search

NUDelta / orchestration-engine

Scripts to guide situated actions, powered by programming constructs that model ways of working
0 stars 0 forks source link

separate out detection and feedback components from script to allow flexibility in their use #8

Open kapil1garg opened 2 years ago

kapil1garg commented 2 years ago

Scripts currently include a detector for when they should run and actionable_feedback that should be run when the detector condition is satisfied in whatever venue is encoded by the actionable_feedback.

In refection from status update and mysore this week, I think there are two usability issues with this approach:

  1. scripts become potentially overly prescriptive where the one feedback that is included with the script must be run when the script detector is triggered
  2. detection and feedback components are reusable. the script as a whole is, but if someone wants to use the detection condition or feedback elsewhere, they effectively need to re-write the script.

Two user stories where this breaks:

  1. A script that suggests talking about resolving ways of overcommitting on a sprint only suggests talking to one's mentor, whereas other venues such as an onboarding mentor (if first quarter student), pairing session on planning during Pair Research, or ad-hoc pairing with a SIG-mate are all viable alternatives that could be used.
  2. A mentor in a SIG meeting wants their student to do learning module during a mysore, and would like to use a pre-made actionable feedback that suggests the activity to the student during the next Studio meeting.

I think the general structure of a script can stay the same, but instead of encoding the detector and actionable_feedback directly in the script, these are actually separate, reusable components (available as "library" functions) that script writers can choose to use if they are relevant to the situations/activities they are trying to support