Specific selection of methods depending on the class object size (scalar or array)

Hello community.

I am developping a class with quite a few methods. Some of these methods are useful when the class object is an array, for example to filter out individual objects that match certain conditions. Some of these methods only make sense when the object is a scalar.

I am trying to implement different behavior depending on the class object size.

I have tried a few things. First defining a parent class for array objects

classdef arrayclass < handle

    properties
        prop1
    end

    methods
        % constructor
        function obj = arrayclass(n)
            obj.prop1 = 1; % scalar object
            if n >1 % or array object
                obj = [obj, obj];
            end
            if isscalar(obj) % if scalar object, try to build children class
                obj = scalarclass(obj);
            end
        end
    end

    methods % shared methods between child and superclass

    end

    methods (Access = ?arrayClass) % methods exclusive to superclass
        function arrayMethod(obj)
            disp(obj.prop1 + 1)
        end
    end
end

and a child class for scalar objects

classdef scalarclass < arrayclass & matlab.mixin.Scalar

    methods % methods exclusive to child
        function obj = scalarclass(arrayObj) % constructor
            obj.prop1 = arrayObj.prop1;
        end
        
        function scalarMethod(obj)
            disp(obj.prop1)
        end
    end
end

I ran into some issues:

• The scalar class calls the array class constructor and I run into issues. or infinite loop
• Even if that would work, I don't know how to update the class type whenever the size changes:

a = arrayClass(2);
a(1).scalarMethod % wont be available since a is detected as arrayClass

Would you have any suggestion on how to proceed ?

Answer 1

[Satwik]

Hi Guillaume,

I understand that you want to create a class that has different behaviour based on whether the object is scalar or vector.

Your implementation of making a parent class for vector and then a child class for scalar methods can be reduced to just as single class by using function isscalar to check whether object is a scalar or vector.

For example if you call filter function and your object is a scalar then just just check with the isscalar function and return an error.

This single class implementation will be a lot easier to read and maintain.

Hope this helps!

isscalar documentaion for more details.