feature: CtLambda#getMethod() and fix of #1159

[pvojtechovsky]

This PR fixes issue #1159. The issue #1159 is caused by wrong return type of CtExecutableReference created for lambda. Therefore this code failed too:

assertTrue(lambdaRef.getDeclaringExecutable().getType()
     .equals(paramRef.getDeclaringExecutable().getType()));

Now there is a question whether solution of this PR is correct.

In current spoon model the interface CtLambda represents both:

• CtExpression, which returns an instance, which implements a type defined by result of CtLambda.getType()
• CtExecutable, which represents an executable element. It means it has parameters, body, thrownTypes ... and return type

The problem was that return type of executable element was not accessible. Therefore during creation of CtExecutableReference of lambda expression, there was used wrong return type of that executable (see change in ExecutableFactory).

Now the issue #1159 is fixed - the parameter references are equal.
... but I have found another issue #1163, which causes that CtLambdaImpl#getExecutableType actually does not work correctly.

[monperrus]

I'd say there is a solution without introducing a new method.

What about setting the type to the correct value when it's available or simply overriding getType in CtLambdaImpl.

[pvojtechovsky]

CtLambdaImpl#getType is already used for different purpose and makes sense too. It cannot be overriden.

What about setting the type to the correct value when it's available

I do not understand. Do you mean to have not derived property CtLambdaImpl#executableType, which is set during compilation of model?

[pvojtechovsky]

an idea: to replace CtLambda#getExecutableType(), by
A) ctLambda.getExecutable().getType()
B) ctLambda.getExecutableMethod().getType()
C) ctLambda.getMethod().getType()
Note: all these methods are derived From the result of ctLambda.getType().map(new SAMMethodFilter()).list() ...

I suggest that, because the new method getExecutableType is too special. More generic method which returns CtMethod, which is implemented by result of lambda expression, is better.

[pvojtechovsky]

I replaced CtLambda#getExecutableType() by CtLambda#getMethod(), which is more useful.

If somebody needs the getExecutableType, then s/he can call ctLambda.getMethod().getType(). It is the same.

I think this code makes sense. The client needs to know, which method is called by lambda expression.

What about the method name CtLambda#getMethod(). Is it good name, or you have better one?

Note: the tests fails because fix of #1163 is not part of this PR (to keep PR simple). After #1165 is merged this PR will pass tests too.

[spoon-bot]

Revapi Analysis results

Old API: fr.inria.gforge.spoon:spoon-core:jar:5.6.0-20170204.234700-25

New API: fr.inria.gforge.spoon:spoon-core:jar:5.6.0-SNAPSHOT

Detected changes: 2.

Change 1

Name	Element
Old	method void spoon.support.visitor.java.reflect.RtMethod::(java.lang.Class, java.lang.String, java.lang.Class, java.lang.reflect.TypeVariable<java.lang.reflect.Method>[], java.lang.Class[], java.lang.Class[], int, java.lang.annotation.Annotation[], java.lang.annotation.Annotation[][], boolean)
New	method void spoon.support.visitor.java.reflect.RtMethod::(java.lang.Class, java.lang.String, java.lang.Class, java.lang.reflect.TypeVariable<java.lang.reflect.Method>[], java.lang.Class[], java.lang.Class[], int, java.lang.annotation.Annotation[], java.lang.annotation.Annotation[][], boolean, boolean)
Code	java.method.numberOfParametersChanged
Description	The number of parameters of the method have changed.
Breaking	binary: breaking, source: breaking

Change 2

Name	Element
Old	none
New	method spoon.reflect.declaration.CtMethod spoon.reflect.code.CtLambda::getMethod()
Code	java.method.addedToInterface
Description	Method was added to an interface.
Breaking	semantic: potentially_breaking, binary: non_breaking, source: breaking

[monperrus]

@return the method that this lambda expression implements. Must be defined as a non-default method in an interface, eg Consumer.accept().

[monperrus]

One question: Is lambda.getMethod() equivalent to lambda.getRefarence().getOverridingExecutable().getDeclaration()?

[pvojtechovsky]

Is lambda.getMethod() equivalent to lambda.getReference().getOverridingExecutable().getDeclaration()?

I made a test and lambda.getReference().getOverridingExecutable() is actually null.

I do not know whether it should be equivalent or not. Probably be yes.

I just know that the expression lambda.getReference().getOverridingExecutable().getDeclaration() cannot be used instead of lambda.getMethod(), because lambda.getMethod() is used internally in lambda.getReference(), so it would end with stack overflow.

[spoon-bot]

Revapi Analysis results

Old API: fr.inria.gforge.spoon:spoon-core:jar:5.6.0-20170205.234548-26

New API: fr.inria.gforge.spoon:spoon-core:jar:5.6.0-SNAPSHOT

Detected changes: 1.

Change 1

Name	Element
Old	none
New	method spoon.reflect.declaration.CtMethod spoon.reflect.code.CtLambda::getMethod()
Code	java.method.addedToInterface
Description	Method was added to an interface.
Breaking	source: breaking, semantic: potentially_breaking, binary: non_breaking

[monperrus]

OK. This really makes sense, and I'm fine with adding this method to lambda, but with a more specific name.

Do you prefer getOverridenMethodor getImplementedMethod or getImplementedInterfaceMethodor something similar?

[pvojtechovsky]

I like both getOverridenMethod or getImplementedMethod.
getOverridenMethod is probably better, because it's meaning is same like CtExecutable#getOverridenMethod. I will rename it

[monperrus]

OK, good.

I did a spelling mistake: getOverriddenMethod with double r and double d

[spoon-bot]

Revapi Analysis results

Old API: fr.inria.gforge.spoon:spoon-core:jar:5.6.0-20170206.234614-27

New API: fr.inria.gforge.spoon:spoon-core:jar:5.6.0-SNAPSHOT

Detected changes: 1.

Change 1

Name	Element
Old	none
New	method spoon.reflect.declaration.CtMethod spoon.reflect.code.CtLambda::getOverriddenMethod()
Code	java.method.addedToInterface
Description	Method was added to an interface.
Breaking	semantic: potentially_breaking, source: breaking, binary: non_breaking