CtType#getAllMethods contract seems not respected

[surli]

Looking on CtType#getAllMethods we have the following comment (see:

spoon/src/main/java/spoon/reflect/declaration/CtType.java

Line 211 in 40e15b2

* Up to 5.2: The recursion stops when the super-type/super-interface is not in the model,

)

Up to 5.2: The recursion stops when the super-type/super-interface is not in the model,
which means that no method of library super-classes, or of Object are present.

However when looking on the test CtTypeInformationTest#testGetSuperclass we explicitely try to obtain methods from Object
See:

spoon/src/test/java/spoon/reflect/declaration/CtTypeInformationTest.java

Line 139 in 40e15b2

// + 12 of java.lang.Object

So either the contract is wrongly defined, or the implementation is wrong.

[monperrus]

The documentation is wrong. (the behavior has been changed with the introduction of shadow classes some time ago)

[surli]

Ok, and do we agree that for a given class and its associated CtClass we should have the following contract:

MyClass.class.getMethods().length == MyCtClass.getAllMethods().size() ?

See: https://docs.oracle.com/javase/7/docs/api/java/lang/Class.html#getMethods()

[monperrus]

No, because we have the methods of objects MyCtClass.getAllMethods().size()  == MyClass.class.getMethods().length + Object.methods.size

[surli]

No, because we have the methods of objects

Actually getMethods() also return Object method. The main difference is getMethods() of Java only return public methods, and do not duplicate overriden method even when they don't have exactly the same signature (like with generics)

[surli]

Fixed in #1644