I came across a use case where sadly I needed to use this approach to support arbitrary variable names. Any idea on how to support tab completion within the parentheses? e.g.:
c.(‘ %tab complete after the quote

Thanks,
Jim

obj = myclass(x);
obj. % this shows the property names that my overloaded properties method returns
i = [1,3];
obj(i). % this shows a popup message saying no completions found. Do you know if that is considered correct/normal Matlab behavior?

I also tried overloading the fieldnames method as well but I still get the no completions found message. Any ideas of how I can get this to work?

Being I am already overloading the subsasgn and subsref methods I do realize I could change them such so that instead of doing obj(i).property_name I could use obj.property_name(i) and then I think the tab completion would work. However, my custom class (obj) actually contains an array of a different custom class. So it seems more natural to specify the objects I want first, obj(i) and then the property to get/set, obj(i).property_name.

In 2013a and 2013b, I am able to hide handles’ methods by inheriting from a class that overloading the functions (addlistener.m, etc) and gives them access of ‘Hidden’. As seen in the stackoverflow questions below:
http://stackoverflow.com/questions/6621850/is-it-possible-to-hide-the-methods-inherited-from-the-handle-class-in-matlab

My question is: Is anyone aware of changes made to R2014a+ that would not allow overloading of ‘methods’ and thus not resolve the hiding of handles’ methods when using the tab-completion tool tip?

I implemented a MEX-function “my_setfield.cpp” to illustrate: http://pastebin.com/j69kQEur
To keep it simple, I only handle scalar structures, but you can extend the code to work with struct arrays..

Here is an example showing the use of illegal field names:

% create struct
s = struct();
s = my_setfield(s, 'a a', 1);
s = my_setfield(s, 'b.b', 2);
s = my_setfield(s, 'c-c', 3);
s = my_setfield(s, '1dd', 4);
s = my_setfield(s, repmat('e',1,70), 5);
s = my_setfield(s, ' asd f @#$^$%&', 6);
 
% retrieve fields using "dynamic field names" syntax
disp(s)
a = s.('a a')
b = s.('b.b')
c = s.('c-c')
d = s.('1dd')
%e = s.(repmat('e',1,70))   % this one didn't work :)
f = s.(' asd f @#$^$%&')

(Note: you can workaround the “namelengthmax” restriction by similarly implementing a C++ MEX-function that uses mxGetField/mxGetFieldByNumber).

I’d go for a consulting project helping them to get rid of that mess :>