MATLAB: ‘Matlab Coder’ generated code is NEVER of double type (only real_T)

Question

I'm trying to convert a piece of matlab code into C++ code. I have one large vector (100+ elements) as input and one large vector as the output. Now, when I try to build the code using the Matlab Coder, while explicitly specifying that the input variable (u) is a DOUBLE, Matlab just seems to ignore my demands and does its own thing by creating the function with input variable of the type const real_T type instead.

This would not be so bad except that I need to use the resulting code with the Legacy_Code function in order to create an S Function. What can I do?

[Attached screenshot shows what is wrong]

Answer 1

See http://www.mathworks.com/matlabcentral/answers/102245

and see the "19" answer at http://stackoverflow.com/questions/2386772/difference-between-float-and-double.

Notice that if you do not include the include file indicated there, you could go ahead with the legacy code tool and create the equivalence: http://www.mathworks.com/help/simulink/slref/legacy_code.html

legacy_code('compile', specs, compilerOptions)

example:

legacy_code('compile', s,...
      {'-DCOMPILE_VALUE1=1', '-DCOMPILE_VALUE2=2',...
       '-DCOMPILE_VALUE3=3'});

So you could '-Dreal_T=double'