Function handle restrictions for MATLAB Coder

Hi,

I read in the mathworks documentation (regarding MATLAB Coder) that:

Do not use the same bound variable to reference different function handles

In some cases, using the same bound variable to reference different function handles causes a compile-time error. For example, this code does not compile:

    function y = foo(p)
    x = @plus;
    if p
      x = @minus;
    end
    y = x(1, 2);

I was wondering what is meant by "in some cases". Are there specific conditions for which it is allowed to store different function handles in the same variable? The reason I am asking is that I am trying to create function that selects and returns a function handle based on some criteria.

    function fcnHandle = select_function(criteria, function_name)
        if criteria == 1
            fcnHandle = str2func(['Package1', '.', function_name]);
        elseif criteria == 2
            fcnHandle = str2func(['Package2', '.', function_name]);
        end
    end

I have 2 packages (Package1 and Package2) and they both contain a function with the exact same function signature (e.g. my_func(a,b)). So, if criteria == 1 i get the function handle for my_func(a,b) from Package1, and criteria == 2 returns the function handle for my_func(a,b) from Package2. This works fine in MATLAB, but when i try to generate code (using MATLAB Coder) i get an error saying:

Type mismatch: MATLAB Coder cannot determine the equivalence of function handles Package1.my_func versus Package2.my_func.

So my question is; can I get around this problem in an easy way? The documentation states that I am not allowed to store different function handles in the same variable "in some cases", so maybe my "case" just needs some tweaking? If it is not possible, I would highly appreciate any suggestions on how to achieve function selection in an alternative manner.

Thanks in advance!

[Walter Roberson]

I wonder what would happen if you used

    function fcnHandle = select_function(criteria, function_name)
        fcnHandles = {str2func(['Package1', '.', function_name]), str2func(['Package2', '.', function_name]) };
        fcnHandle = fcnHandles{criteria};
    end

or if you used

        function fcnHandle = select_function(criteria, function_name)
          fcnHandle = str2func(sprintf('Package%d.%s', criteria, function_name));
        end

[Øystein Henriksen]

Thanks for your suggestion!

I tried creating a cell array of function handles and returning the correct cell based on my criteria, but I get the same error. I think the issue is that the function handles refer to different functions, and thus cannot be referenced by the same variable. This will also be an issue (when using MATLAB Coder) with your second suggestion.

Answer 1

[Chidvi Modala]

The example from the documentation shows a single variable being reassigned to a different function handle. However, the other case refers to using cell arrays. Since cell arrays support data of different types, assigning each element of a cell array to a different function handle doesn't give a compile time error. You can make use of the following example

function out = foo(p)
x = cell(2,1);
x{1} = @minus;
x{2} = @plus;
if p
    out = x{1}(1,2);
else
    out = x{2}(1,2);
end
end

[Øystein Henriksen]

Thanks for your answer.

In this case, the variable out will be the result of either a minus operation or a plus operation. Anyway, the out-variable will contain a double. What I want to do is to return the actual function handle for (in this example) either the plus- or the minus function.

[Walter Roberson]

I would be concerned about whether you would be violating " Do not pass function handles to or from entry-point functions" ?

[Øystein Henriksen]

I would be calling the get_function_handle-function inside all my entry-point functions. The structure would be like this:

function entry_point_func_1(something)

<code>

fcnHandle = get_function_handle(someCriteria)

someVariable = fcnHandle(someInput)

<code>

end

The function "get_function_handle" would itself not be an entry point function.