Skip to main content

Adding a new calibration type

info

This is subject to change! Currently we don't like how much a developer has to implement (or copy and paste) to create a new calibration. We want to make it easier to design new calibrations.

Ideally, all calibrations, including od_calibration and pump_calibration, should behave similarly:

  1. Interface should have the following:

    • pio run <x>_calibration starts the calibration and saves it keyed by a unique name (see 2. for storage)
    • pio run <x>_calibration list lists all saved calibrations, keyed by their unique name.
    • pio run <x>_calibration display ?name? displays information about the current calibration to be used, or the calibration ?name? if provided
    • pio run <x>_calibration change_current <name> changes the current calibration to <name> calibration.
    • pio run <x>_calibration publish <name> publishes the calibration to the leader.

  2. On disk, all run calibrations should be stored in local persistent storage under <x>_calibrations keyed by a unique name, and the current calibration should be stored in <x>_current_calibration, with appropriate key (can use <calibration_type>) that is not the unique name, but something consistent. Note: The name cannot be current.

  3. A struct should be created / sub-classed from structs.Calibration that will encode / decode the calibration data blob.

  4. When a new calibration is created, a PUT request to /api/calibrations/ should be sent. The body is the json-encoded Calibration struct.

  5. When a new calibration is set as current (change_current), a PATCH request to /api/calibrations/<pioreactor_unit>/<calibration_type>/<calibration_name> should be sent.

  6. Creating a new calibration should both publish to leader and set as current.

For example, for pump_calibration, the three default calibration types are: media_pump_calibration, alt_media_pump_calibration, and waste_pump_calibration.