findbugs:NP_NULL_ON_SOME_PATH_FROM_RETURN_VALUE does not consider null validation

[alixwar]

Problem: A violation is triggered at the last line, saying that args may be null. This is clearly validated the line before (using org.apache.commons.lang3.Validate)

Affected version: 3.1.2 (SpotBugs)
Seen in: SonarQube (FindBugs plugin 3.7)

Reproducible code example (from an Android app):

@Override
public void onCreate(@Nullable Bundle savedInstanceState) {
   super.onCreate(savedInstanceState);
   final Bundle args = getArguments();
   Validate.notNull(args);
   email = args.getString(ARG_EMAIL);
}

[npfistner]

Problem: Also if validated using java assert spotbugs still complains NP_NULL_ON_SOME_PATH_FROM_RETURN_VALUE

Affected version: 3.1.5 (SpotBugs)

        assert entity.getId() != null;
        final long id = entity.getId().longValue();

[ThrawnCA]

I'm not sure that the assert keyword should be relied on, since assertions can be disabled. But the Commons Lang one is likely worth including; thanks for pointing it out.

[marcelmore]

Problem: Same problem with Google Guava

Namely on the two checks:

com.google.common.base.Verify.verifyNotNull
com.google.common.base.Preconditions.checkNotNull

Seen in: Gradle build with SpotBugs plugin:

id "com.github.spotbugs" version "1.6.9"

Affected Version: 3.1.12

[alixwar]

Same issue with java.util.Objects.requireNonNull()

[nmatt]

+1 for supporting requireNonNull().

[gmbett]

Same issue with custom method that is annotated with @contract("null -> false")

[alixwar]

I'm trying to add a reproducer in the SpotBugs project but I keep getting:

The following classes needed for analysis were missing:
  org.apache.commons.lang.Validate

even after adding the dependency to the spotbugs-tests module 😕

[gmbett]

Any updates on this issue?

[victornoel]

@ThrawnCA any news on that?

[C-Otto]

There's a database of hardcoded method parameter annotations in https://github.com/spotbugs/spotbugs/blob/master/spotbugs/src/main/java/edu/umd/cs/findbugs/ba/DefaultNullnessAnnotations.java#L406

At the given line, the first argument of Preconditions.checkNotNull is assumed to be "non-null". As such, the analysis complains if null is passed to checkNotNull.

The same holds true for other implementations, for example Objects.requireNonNull.

[ByeBug]

any news on that and #456 ?

[hazendaz]

Fixed by #2709