TUnit

A unit test framework for Objective-C.

What for? – History

Inspired by TDD talks on JAOO 2003 I found that I am lacking an Objective-C unit test framework.

Searching the net I found OCUnit and some other I forgot about. Both were either bound to Apples Cocoa Framework or did not behave like I expected. So I decided to write my own one on top of my own little Foundation (TFoundation - not public as of September 2010).

Later I missed mock objects and build them in inspired by EasyMock.

Then I found more inspiration when testing Ruby code with RSpec and I added a better mocking framework based on dynamic modification of objects and classes.

Requirements

TUnit actually requires the non public TFoundation. Nevertheless it would run with any other “Foundation” if one implements a tiny wrapper layer (which I actually have done for the very old libFoundation and which should be easy to transfer to Cocoa or GNUStep).

TUnit requires the usage of “-fobjc-exceptions” (@try/@catch…).

TUnit was build with and for the GNU runtime. It will not run on the Apple runtime without modifications.

TUnit may not run well (or at all ;)) on non x86/x64 systems or on non linux boxes.

Why on GitHub?

So, if the stuff is so useless, why it's on GitHub?

Because its my private project which I am using in my professional environment by cloning it as a submodule.

How to use it anyway?

First, you could contact me via GitHub and ask for

• a public TFoundation
• support for Cocoa, GNUStep or whatever
• support for the Apple runtime
• a wrapper layer to run on Cocoa or GNUStep …
• help on compiling the stuff
• any other Objective-C/TUnit related stuff

Second, wait for my reply. I will not do any of the above without needing it for myself or someone asking for it. And don't hesitate: One question could be enough.

How to use it when compiled and (propably) running?

• All test classes inherit from TTestCase.
• All test classes ending with TestCase will not be run.
• The instance method -(void)setUp is performed before every test and the instance method -(void)tearDown is performed after every test.
• Every instance method beginning with test or itShould is performed as a single test.
• Assertions are made by macros (to get file and line of the assertion on failure) like ASSERT(…) or ASSERTEQUALS(…, …).
• A test run can be limited to classes or methods.

Mocking

Mocking is easy as:

id myObj = [MyClass new];
[[[myObj shouldReceive] myMessage: someArg] andReturn: theResult];
[myObj methodUnderTest];

If myMessage: is not being called with someArg the test will fail.

Stubbing is also available:

[[[myObj stub] myMessage: someArg] andReturn: theResult];

Also some more tweaks are available (skip checking of some/all stubbed/mocked method parameters; ordered messages; message call count). I will document all of it if someone performed the first step of “How to use it anyway?”.

Any questions?

As mentioned in “How to use it anyway?”, contact me via GitHub.