I had make a last minute change, as I was using (and improving) the meas_base Centroid and ShapeUtilities, without thinking that I couldn't actually use them in afw unit tests.
I fixed some of the problem using the Point2D and Quadrupole functor keys, but there is not really a great solution for the errors. The CentroidResult, ShapeResult and their functor keys are really needed to do comparisons of the Slot error results and the results read directly from the fields. I cobbled together a solution for the testSourceTable tests, but it was not as clean as using the ResultKeys from meas_base. They allow the user to get the error values back as a CovarianceMatrix, no matter how they are stored in the table. I think this capability should be in afw, not in meas_base, but I will let Jim figure this out.
I will not seek another review, but if anyone wants to do one, I will not merge until tomorrow night. Changes are only in afw now. u/pgee/DM-1076
I had make a last minute change, as I was using (and improving) the meas_base Centroid and ShapeUtilities, without thinking that I couldn't actually use them in afw unit tests.
I fixed some of the problem using the Point2D and Quadrupole functor keys, but there is not really a great solution for the errors. The CentroidResult, ShapeResult and their functor keys are really needed to do comparisons of the Slot error results and the results read directly from the fields. I cobbled together a solution for the testSourceTable tests, but it was not as clean as using the ResultKeys from meas_base. They allow the user to get the error values back as a CovarianceMatrix, no matter how they are stored in the table. I think this capability should be in afw, not in meas_base, but I will let Jim figure this out.
I will not seek another review, but if anyone wants to do one, I will not merge until tomorrow night. Changes are only in afw now. u/pgee/
DM-1076