Question about FieldClockModel

The FieldClockModel interface has been added recently during generalization of QuadraticClockModel into PolynomialClockModel.

This interface introduces a getReferenceDate method that has no counterpart in the ClockModel interface. It seems to be used only in PolynomialFieldClockModel, where it makes sense. Indeed, there is a reference date in the regular double PolynomialClockModel.

I would like to move this reference date from the FieldClockModel interface level to the PolynomialFieldClockModel class level.

Would anyone mind?

I think I will go further.
There are now both a getFieldModel (but restricted to Gradient), and getFieldOffset (which accepts any field which look redundent to me. There are also several models do not have a corresponding field (aggregated, sum, difference, sampled). Aggregated delegate to an underlying model but fixes the extraction date, whereas the other ones return an exception when converted to field.
I will try to be more generic (any field) and general (all clock models), and probably rename the generic version of getFieldModel into toField, for consistency with GNSSOrbitalElements.toField design (which I also intend to generalize to ParameterDriver as per issue 2000.

I have open issue 2007 for that.

@brian.carter.he360 do you have any concerns about this change?

Hello, no, I have no concerns with this plan. Thanks for identifying the inconsistencies and addressing them!