Junit and cactus
- 1. Junit and Cactus Dr. Himanshu Hora SRMS College of Engineering & Technology Bareilly (INDIA)
- 2. Table of Contents ď‚— ď‚— ď‚— ď‚— ď‚— ď‚— ď‚— Overview Article Outline Thesis JUnit Background Cactus Background Pitfalls 1-6 Conclusion
- 3. Overview  Junit and Cactus are popular tools for automating      testing of Java classes and web-based components Automation accomplished using ANT (Java Make utility) and often tested every time project built These tools enable testers to “verify” the code is written correctly – it was built right Automated tests are very useful to show code works correctly esp. after refactoring which is central to XP I picked this article because deals with JUnit & Cactus, offers a critique of the tools, and was practical, not just theoretical As this article describes, the tests themselves must be built right in order to validate the code being tested
- 4. Outline ď‚— Theme: Risk when testing with JUnit & Cactus ď‚— Risk 1: No assert ď‚— Risk 2: Unreasonable assert ď‚— Risk 3: Console-Based Testing ď‚— Risk 4: Unfocused Test Method ď‚— Risk 5: Failure To Isolate Each Test ď‚— Risk 6: Failure to Isolate Subject
- 5. Analysis  Thesis: critique was correct but wordy, repetive, and      incomplete – missing some bigger pitfalls Risk 1-3 are all the same – use assert correctly Risk 4 is programming rule – write focused method Risk 5 is JUnit rule – Use setUp and tearDown Risk 6 is incomplete analysis of Cactus vs MockObj Overview of Junit and Cactus, then discuss pitfalls, and finally look at unmentioned pitfalls & issues
- 6. JUnit Overview  Popular and simple Java framework / library for        automating testing Integrates well with ANT – Java Make utility General idea: write one test class per testee Write one method to verify each main feature Test class must extend TestCase and each test method must start with “test” Order of test method execution varies Use assertTrue() and assertEquals() to verify code Use setUp() & tearDown() prepare testcase testfixture
- 7. JUnit code example Following test cases test the collection methods, isEmpty() and add() import junit.framework.*; public class SimpleTest extends TestCase { private java.uti.Collection collection; protected void setUp() { collection = new ArrayList(); } // instantiates collection test fixture protected void tearDown() { collection.clear(); } public void testEmptyCollection() { assertTrue(collection.isEmpty()); } public void testOneItemCollection() { collection.add("itemA"); assertEquals(1, collection.size()); } }
- 8. Cactus Overview ď‚— Built on Junit framework ď‚— Intended to test JSP, Servlets, EJBs, Filters, and custom tags ď‚— Complex architecture that has client JVM call the J2EE application server JVM via redirector ď‚— Testcase classes must reside on client and server ď‚— Adds two methods to Junit architecture, beginXX() and endXX() which get called on client, rest on server
- 10. Cactus Sequence Diagram – Jakarta website
- 11. Risk 1-3 – No assert, Unreasonable assert, Console-Based Testing  Risk 1-3 are all the same – use assert correctly  Very important and author points out a major guideline  test what is written in the javadocs for the testee  implies javadocs must be up-to-date with requirements  Author also points out to write a test case if encounter a defect before it is corrected  However, using assertTrue() and assertEquals() is obvious  These are the prominent features of the JUnit
- 12. Risk 4 – Unfocused Test Methods  Writing focused tests is really just writing good code  General rule of programming to make methods succinct, this applies equally to test methods  Writing focused test methods is the whole point
- 13. Risk 5 – Failure To Isolate Each Test Risk 5 is really saying to use setUp() and tearDown() to prepare/release test fixture, an obvious suggestion – example from JUnit site Bigger pitfall is automating creation of the test fixture in distributed environments import junit.framework.*; public class SimpleTest extends TestCase { private java.uti.Collection collection; protected void setUp() { collection = new ArrayList(); } // instantiates collection test fixture for 2 tests protected void tearDown() { collection.clear(); } public void testEmptyCollection() { assertTrue(collection.isEmpty()); } public void testOneItemCollection() { collection.add("itemA"); assertEquals(1, collection.size()); } }
- 14. Risk 6 – Failure to Isolate Subject  Author points out one drawback of Cactus is that does not isolate test case as MockObjects does  MockObjects simulates the Servlet container  Mock Object framework does isolate test but at big expense  Massive amount of stubs needed, more code to maintain  While Cactus may be better than MockObjects, it may NOT be better than HttpUnit, why not compare these?
- 15. Alternative tools ď‚— HttpUnit cleaner, simpler tool than Cactus ď‚— HttpUnit is black box testing by calling webserver ď‚— Test code resides ONLY on client JVM ď‚— Various interfaces like JWebUnit (Java API) and ď‚— ď‚— ď‚— ď‚— WebTest (XML) integrate well with ANT Use Junit for unit tests and HttpUnit for functional Features to analyze HTML, ie. table element tests Features to input HTML form elements http://www.junit.org/news/extension/index.htm
- 16. Conclusion ď‚— Article lists some useful guidelines & pitfalls in an wordy fashion ď‚— Many pitfalls were obvious and important ones not mentioned ď‚— Important pitfalls not mentioned include ď‚— ď‚— ď‚— ď‚— ď‚— ď‚— Cost, complexity, difficulty of distributed tests not mentioned Performs white box tests, yet, JUnit already does this Does not test HTTP interface (tests presentation layer poorly) Test code must reside in same package as testee & both JVMs Testers must be programmers JWebUnit & WebTest better for web unit testing ď‚— At times unclear when addressing Junit vs. Cactus and unnecessarily complex coding examples ď‚— However, automating testing can save time and money in the long run ď‚— These tools, while not perfect, are major players for automated Java testing and can verify functionality during development and refactoring
- 17. THANK YOU Dr. Himanshu Hora SRMS College of Engineering & Technology Bareilly (INDIA)