Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 485302

Summary:	[1.8][null]Missing warnings when using Wildcards with null annotations
Product:	[Eclipse Project] JDT	Reporter:	Till Brychcy <register.eclipse>
Component:	Core	Assignee:	Till Brychcy <register.eclipse>
Status:	VERIFIED FIXED	QA Contact:	Stephan Herrmann <stephan.herrmann>
Severity:	normal
Priority:	P3	CC:	dusisarath, manoj.palat, stephan.herrmann
Version:	4.5.1
Target Milestone:	4.6 M6
Hardware:	PC
OS:	Mac OS X
See Also:	https://git.eclipse.org/r/63686 https://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=e2fc6e7db19ada9b01733b33b99936b7344de6b1
Whiteboard:

Description Till Brychcy

2016-01-06 18:20:23 EST

Comment 1 Till Brychcy

2016-01-06 18:25:27 EST

DESCRIPTION:
During compilation of the following code, only "warning 3" is reported:

import org.eclipse.jdt.annotation.NonNull;
import org.eclipse.jdt.annotation.Nullable;

public class WildCardNullable {
	static class A<T> {
		@Nullable
		T returnNull() {
			return null;
		}

		void acceptNonNullT(@NonNull T t) {
		}

		void acceptNonNullObject(@NonNull Object x) {
		}
	}

	static @NonNull Number g(A<? extends @NonNull Number> a) {
		return a.returnNull(); // error 1 expected
	}

	public static final <T> void map(final A<@NonNull ? super T> a, T t) {
		a.acceptNonNullT(t); // warning 2 expected
		a.acceptNonNullObject(t); // warning 3 expected
	}
}

ANALYSIS:
In NullAnnotationMatching.areSameTypes, for capture bindings, the null bits are not compared if the bounds were found.

PATCH (via gerrit): 
compare the null bits in these cases. adjust testBug448709 so it expects are warning, too.

Comment 2 Eclipse Genie

2016-01-06 18:28:58 EST

New Gerrit change created: https://git.eclipse.org/r/63686

Comment 3 Till Brychcy

2016-01-27 15:09:10 EST

This didn't have Target Milestone at all, feel free to correct it if you plan to look at this later.

Comment 4 Stephan Herrmann

2016-02-04 12:06:48 EST

Looks good to me.

Looking at the change in testBug448709() I only added one more (stripped-down) variant of that test, witnessing that the new warning is *not* raised, when flow analysis knows that the value is nonnull.

Rebased, test added and re-uploaded to gerrit.

Comment 5 Eclipse Genie

2016-02-04 15:02:50 EST

Gerrit change https://git.eclipse.org/r/63686 was merged to [master].
Commit: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=e2fc6e7db19ada9b01733b33b99936b7344de6b1

Comment 6 Stephan Herrmann

2016-02-04 15:04:34 EST

(In reply to Eclipse Genie from comment #5)
> Gerrit change https://git.eclipse.org/r/63686 was merged to [master].
> Commit:
> http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=e2fc6e7db19ada9b01733b33b99936b7344de6b1
> 

Released for 4.6 M6. Thanks!

Comment 7 Dusi Sarath Chandra

2016-03-16 06:50:41 EDT

Didn't find UI option for pessimistic null analysis for free type variables. After manually adding the same I could reproduce the mentioned scenario in comment 1.

Verified for 4.6M6 using Version: Neon (4.6)
Build id: I20160314-2000

Comment 8 Stephan Herrmann

2016-03-18 12:53:18 EDT

(In reply to Dusi Sarath Chandra from comment #7)
> Didn't find UI option for pessimistic null analysis for free type variables.

I assume this got resolved by using a more recent build, right?