The following code compiles fine in IntelliJ and Eclipse, but the JDK compiler 1.8.0_25 complains. First, the code.
import java.util.function.Predicate;
public abstract class MyStream<E> {
static <T> MyStream<T> create() {
return null;
}
abstract MyStream<E> filter(MyPredicate<? super E> predicate);
public interface MyPredicate<T> extends Predicate<T> {
@Override
boolean test(T t);
}
public void demo() {
MyStream.<Boolean> create().filter(b -> b);
MyStream.<String> create().filter(s -> s != null);
}
}
The output from javac 1.8.0_25 is:
MyStream.java:18: error: incompatible types: incompatible parameter types in lambda expression
MyStream.<Boolean> create().filter(b -> b);
^
MyStream.java:18: error: incompatible types: bad return type in lambda expression
MyStream.<Boolean> create().filter(b -> b);
^
? super Boolean cannot be converted to boolean
MyStream.java:19: error: bad operand types for binary operator '!='
MyStream.<String> create().filter(s -> s != null);
^
first type: ? super String
second type: <null>
MyStream.java:19: error: incompatible types: incompatible parameter types in lambda expression
MyStream.<String> create().filter(s -> s != null);
^
Note: Some messages have been simplified; recompile with -Xdiags:verbose to get full output
4 errors
When I replace ? super E
with simply E
, JDK compiles successfully.
When I replace filter(MyPredicate
with filter(Predicate
, JDK compiles successfully.
Since it works with JDK 1.8.0_60, I suspect it is a compiler bug.
Any details on what caused this and when it has been fixed?
A return statement is not an expression in a lambda expression. We must enclose statements in braces ({}). However, we do not have to enclose a void method invocation in braces. The return type of a method in which lambda expression used in a return statement must be a functional interface.
It depends. Whenever you find yourself using the same lambda in different places you should consider implementing a class that implements the interface. But if you would've used an anonymous inner class otherwise I think a lambda is far better.
The lambda expression must return a result that is compatible with the result of the functional interface method. If the result is void , the lambda body is void-compatible. If a value is returned, the lambda body is value-compatible.
A lambda expression is a short block of code which takes in parameters and returns a value. Lambda expressions are similar to methods, but they do not need a name and they can be implemented right in the body of a method.
If a lambda expression appears in a target type with wildcards (as in most cases)
Consumer<? super Boolean> consumer = b->{...}
the question arises - what's the type of the lambda expression; in particular, the type of b
.
Of course, there could be many choices due to the wildcards; e.g. we could explicitly choose
Consumer<? super Boolean> consumer = (Object b)->{...}
However, implicitly, b
should be inferred as Boolean
. This makes sense since the consumer should only be fed with Boolean
anyway.
http://docs.oracle.com/javase/specs/jls/se8/html/jls-15.html#jls-15.27.3
If T is a wildcard-parameterized functional interface type and the lambda expression is implicitly typed, then the ground target type is the non-wildcard parameterization of T
(This probably assumes that wildcards are use properly variance-wise on the target type; we might find some hilarious examples if the assumption doesn't hold)
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With