Job Introduction: Testing Applications Related to Programming Design Using MATLAB

Testing applications related to programming design using MATLAB.

“The MATLAB syntax itself can be up and running in 1-2 weeks, but to make the testing professional, reusable, and embeddable in CI — it requires another 1-2 months to master the testing framework, reporting, and automation routines; overall, it is considered ‘easy to start, moderately difficult to master’.”

Here’s a quick guide to building a deliverable testing system from scratch, just follow the steps directly.

1. Difficulty and Time (Tested Personally)

Table

Copy

Phase	Content	Time Required	Difficulty
① Syntax + Matrices	Variables, scripts, functions, plotting	8-10 h	★☆☆
② Testing Framework	matlab.unittest + mock + parameterized	2-3 days	★★☆
③ Automation	CLI + JUnit XML + Cobertura + CI (Jenkins/GitHub)	3-5 days	★★☆
④ Reporting	HTMLTestReport + PDF + Requirement Traceability	1-2 days	★★☆
⑤ Project Practice	Integrate ①-④ into a real algorithm/APP	1-2 weeks	★★☆

Conclusion: ① Very easy; ②-⑤ have patterns, just copy the templates; overall, it has 2 more pitfalls than Python testing (license + CLI), but the learning curve is still friendly.

2. Three-Step Quick Start (with Commands)

➊ Evening of Day 1: Write the First Function + Unit Test

matlab

Copy

%% Function to be tested power2.m
function y =power2(x)
    y = x.^2;
end

%% Test file test_power2.m (class-based)
classdef test_power2 < matlab.unittest.TestCase
methods(Test)
function test_scalar(testCase)
            act = power2(3);
            testCase.verifyEqual(act, 9);
end
function test_vector(testCase)
            act = power2([123]);
            testCase.verifyEqual(act, [149]);
end
end
end

Run:

bash

Copy

matlab -batch "runtests('test_power2', 'ProduceJUnitReport', true)"
→ Generates power2_test.xml (CI recognizes it directly)

➋ Day 2: Parameterization + Coverage

matlab

Copy

classdef test_power2 < matlab.unittest.TestCase
properties(TestParameter)
        x = {0, 1, -2, [12; 34]};
end
methods(Test, ParameterCombination='exhaustive')
function test_power2_param(testCase, x)
            testCase.verifyEqual(power2(x), x.^2);
end
end
end

Coverage:

bash

Copy

matlab -batch "coverage = codecov(power2); disp(coverage)"
→ Outputs Cobertura XML for SonarQube to read.

➌ Day 3: One-Click HTML Report + CLI Embedded in CI

matlab

Copy

import matlab.unittest.TestRunner;
import matlab.unittest.plugins.HTMLTestReportPlugin;
runner = TestRunner.withTextOutput;
runner.addPlugin(HTMLTestReportPlugin.producing('./report.html'));
result = runner.run(packageFolder);

GitHub Action Snippet:

yaml

Copy

- name: Run MATLAB tests
  uses: matlab-actions/run-tests@v1
  with:
    testResultsJUnit: test-results.xml
    codeCoverageCobertura: coverage.xml
    sourceFolder: src

Open <span>report.html</span> in the evening to see:

• Each test case PASS/FAIL trace

• Coverage heatmap

• Requirement ↔ Test bidirectional trace (if you write the Requirement number in the test properties)

3. Common Pitfalls & Quick Solutions

Table

Copy

Pitfall	Quick Solution
No license	Use MATLAB Online (web version) or apply for a 30-day trial; use <span>-batch</span> in CI to avoid GUI pop-ups.
Want to test Simulink models	Use <span>sltest.testmanager</span> to drag test cases, which can also export JUnit + coverage.
Need to mock external hardware	Use <span>matlab.mock.TestCase</span> framework, or simply <span>assignin('base', 'hwRead', @mockRead)</span><span> to inject mock data.</span>
Need PDF reports	<span>exportHTMLReport</span> then use <span>html2pdf</span> in one command.

4. Advanced Roadmap (1-2 Months)

1. Write requirement keywords into <span>Test</span> tags → automatically generate traceability matrix (IEC-61508/ISO-26262 likes this).

2. Use MATLAB Test Manager to graphically maintain test cases → non-programmers can also add or delete cases.

3. Integrate with Simulink Coverage and Polyspace → achieve a full process of model-code-coverage-static checks.

4. Link to Jenkins/SonarQube → automatically run regression, generate reports, and send emails on each push.

5. Summary in One Sentence

MATLAB testing is “simple syntax, mature framework, beautiful reports,” by following the above 3-day template, you can deliver a CI-reusable automated testing project; subsequently, add requirements, coverage, and safety standards, and you can reach enterprise delivery level in 1-2 months.