Main Content

Why Use Native Cell Arrays and Struct Arrays?

In Java®, there is no direct representation available for MATLAB® struct arrays and cell arrays.

As a result, when an instance of MWStructArray or MWCellArray is converted to a Java native type using the toArray() method, the output is a multi-dimensional Object array which can be difficult to process.

When you use MATLAB Compiler SDK™ packages with RMI, however, you have control over how the server sends the results of MATLAB function calls back to the client. The server can be set to marshal the output to the client as an MWArray (com.mathworks.toolbox.javabuilder package) sub-type or as a Java native data type. The Java native data type representation of MWArray subtypes is obtained by invoking the toArray() method by the server.

Using Java native representations of MATLAB struct and cell arrays is recommended if both of these are true:

  • You have MATLAB functions on a server with MATLAB struct or cell data types as inputs or outputs

  • You do not want or need to install MATLAB Runtime on your client machines

Using Native Types Does Not Require a Client-Side MATLAB Runtime

The classes in the package (in javabuilder.jar) do not need a MATLAB Runtime. The names of the classes in this package are the same as those in com.mathworks.toolbox.javabuilder — allowing the end-user to easily create instances of or that work the same as the like-named classes in com.mathworks.toolbox.javabuilder — on a machine that does not have a MATLAB Runtime.

The availability of a MATLAB Runtime on the client machine dictates how the server should be set for marshaling MATLAB functions, since the MWArray class hierarchy can be used only with a MATLAB Runtime. If the client machine does not have a MATLAB Runtime available, the server returns the output of toArray() for cell or struct arrays as instances of or