Convert Code Containing Structures to Fixed Point

This example shows how to convert a MATLAB^® algorithm containing structures to fixed point.

In a local writable folder, create the function struct_fcn.m

function [out, y] = struct_fcn(in)
    % create a nested struct
    z = struct('prop1', struct('subprop1', 0, 'subprop2', [3 4 45]));
    % copy over the struct
    y = z;
    y.prop1.subprop1 = y.prop1.subprop1 + in;
    out = y.prop1.subprop1;
end

In the same folder, create a test file, struct_fcn_tb.m, that calls the function.
```
for ii = 1:10
    struct_fcn(ii);
end
```
At the command line, create a coder.FixPtConfig object, and specify the test bench as struct_fcn_tb.m.
```
fixptcfg = coder.config('fixpt');
fixptcfg.TestBenchName = 'struct_fcn_tb';
```
Generate fixed-point code for struct_fcn.m
```
fiaccel -float2fixed fixptcfg struct_fcn
```
Since no input types are specified, the fixed-point conversion process simulates the test bench to infer types. To specify input types, use the -args option when you enter the fiaccel command.
The automated fixed-point conversion process uses the test bench to simulate the function, gather range information, and propose data types. After fixed-point conversion is complete, in the command window, click struct_fcn_report.html to inspect the fixed-point data types.
When the names, number, and types of fields of two or more structures match, automated fixed-point conversion proposes a unified type. In this example, the range of z.prop1.subprop1 is [0,0], while the range of y.prop1.subprop1 is [0,10]. The app proposes a data type of numerictype(0,4,0) for both z.prop1.subprop1 and y.prop1.subprop1 based on the union of the ranges of the two fields.
Click View Report to see the project summary details and links to the fixed-point MATLAB code and conversion report.