Dear Watchers – forgive my unceremoniously adding you to this ticket, but I wanted to ask for some design advice and it seemed like posting the following essay to HipChat would be excessive!
HSC has changed the default flux measurement used by PhotoCal from PSF to aperture flux. Per discussion on HipChat, the consensus is that this change should propagate to LSST. It's trivial to do this by simply pointing the (new) CalibFlux slot at a CircularApertureFlux field: base_CircularApertureFlux_4 would be equivalent to the 12 pixel flux.naive used on HSC in the default configuration.
However, that directly ties the CalibFlux to the vagaries of the CircularApertureFlux algorithm configuration. For example, the “4” is a reference to a particular entry in a list of radii; adding another radius to the CircularApertureFlux configuration might unintentionally change the CalibFlux.
An alternative would be to add way to automatically choose the appropriate entry in the list of radii based on a radius specified in the task configuration. But that would tie the CalibFlux into the specifics of CircularApertureFlux implementation: It would be more awkward to simply put the PsfFlux in the slot instead, for example.
Another alternative would be to simply define a NaiveFlux measurement, which would consist of running CircularApertureFlux with a single aperture. Changing the aperture used for NaiveFlux would more directly and obviously couple with the value used for CalibFlux. But it would mean we were effectively running the CircularApertureFlux plugin twice, with different configurations, which seems less than ideal.