Le 01/06/2012 17:43, Gilles Sadowski a écrit : > Hi. > > I think that "PolynomialFitter" class deserves to be deleted. > > It looks like a bad use of inheritance because it cannot prevent bad usage > of its base class's "fit" methods (where a user can provide a parametric > function argument that is not a polynomial). > > Issue 798 could be solved by providing the methods with the "maxEval" and > "initialGuess" arguments but this would not solve the above problem. > > The alternative would be to use composition (i.e. have "CurveFitter" an > instance variable of "PolynomialFitter") but this would mean to duplicate > all the methods of "CurveFitter". > I don't think that it is reasonable; all that would gained is saving the > typing of > new PolynomialFunction.Parametric() > in the call to the "fit" method (from "CurveFitter"). > > I propose do deprecate "PolynomialFitter" in 3.1, and remove it in 4.0. > > [The fitting of a polynomial using a "CurveFitter" directly could be > described in the userguide.]
OK from me if this is described in the user guide. Luc > > > Best, > Gilles > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
