ffc8dad1ec
Wrench is not happy, but jenkins bots are fine. This is likely something internal to macOS which varies per version. |
||
|---|---|---|
| .. | ||
| Mac | ||
| iOS | ||
| ApiBaseTest.cs | ||
| ApiCMAttachmentTest.cs | ||
| ApiClassPtrTest.cs | ||
| ApiCoreImageFiltersTest.cs | ||
| ApiCtorInitTest.cs | ||
| ApiFieldTest.cs | ||
| ApiPInvokeTest.cs | ||
| ApiProtocolTest.cs | ||
| ApiSelectorTest.cs | ||
| ApiSignatureTest.cs | ||
| ApiStructTest.cs | ||
| ApiTypoTest.cs | ||
| ApiWeakPropertyTest.cs | ||
| CoreSelectorTest.cs | ||
| EnvironmentVariable.cs | ||
| README.md | ||
| xamarin1.png | ||
README.md
Introspection Tests
Introspection tests are executed on target (both simulator and device for iOS) or a specific version of OSX. The application proceed to analyze itself using:
System.Reflectionfor managed code; and- the ObjectiveC runtime library for native code
and compare the results. E.g. if using .NET reflection it can see a binding
for a NSBundle type then it should be able to find a native NSBundle
type using the ObjC runtime functions. Otherwise an error is raised...
Since the application analyze itself it must contains everything we wish to test. That's why the introspection tests needs to be built with the managed linker disable, i.e. "Don't link".
Pros
- The tests always tell the truth, which can differ from documentation or header files;
Cons
- Incomplete - Not everything is encoded in the metadata / executable;
- Too complete - Not every truth is good to be known (or published), which requires creating special cases in the tests