QFunctions try to recycle as much of the QDataSet interface as possible to define functions. Functions take N parameters as input and result in M parameter output. The N parameters are passed into value as a rank 1 bundle QDataSet, or rank 0 dataset when there is just one input. The M parameter output is returned in a rank 1 bundle dataset. The method exampleInput returns an example input that allows for discovery of the function. Implementations will generally extend AbstractQFunction, which implements values() and exampleOutput(). Goals:
Discover an example input. Result is a rank 1 bundle QDataSet.
slice(0) is the first argument, slice(1) would be the second, etc. This would be a bundle. Note, for functions that have only one argument, like F(T)→[R,MLT,MLAT], this may return a rank 0 dataset. Clients should pass a dataset to the value method a dataset with the same geometry.QFunction ff= TestFunction(); ff.exampleInput().length(); // how many parameters the function takes QDataSet bds= ff.exampleInput().property( QDataSet.BUNDLE_0 ); bds.slice(0).property( QDataSet.UNITS ) // function should handle convertible units (e.g. TimeAxes Ephemeris). bds.slice(0).property( QDataSet.VALID_MIN ) // absolute limits of domain of the function bds.slice(0).property( QDataSet.VALID_MAX ) bds.slice(0).property( QDataSet.TYPICAL_MIN ) // domain of the function parameter bds.slice(0).property( QDataSet.TYPICAL_MAX ) bds.slice(0).property( QDataSet.CADENCE ) // granularity of the function parameter bds.slice(0).property( QDataSet.LABEL ) // label for the parameter
Discover an example of output. Result is a rank 1 bundle QDataSet. This was introduced to support QFunctions where it would be expensive to calculate an input that would result in a meaningful output. It's assumed that many implementations will simply be:
value( exampleInput() );