@Pointcut vs. S1186 & S2325

[sandro-h (Sandro Heinzelmann)]

Hi there,

One of our project has started using AspectJ’s @Pointcut annotation. Eg:

@Pointcut("within(com.package.Service+)")
private void isMyService() {}

Such methods tend to be empty, so naturally Sonar complains about:

• Methods should not be empty (S1186)
• “private” methods that don’t access instance data should be “static” (S2325)

I know we have multiple options to deal with this: for S1186 we can add a comment which might anyway help point out what the empty method is for. For S2325 we could simply disable the rule in such cases.

I was wondering however if you have considered handling such AOP cases and ignoring the empty method.
It seems this was done for the “unused methods” rule in the past: https://groups.google.com/forum/#!searchin/sonarqube/pointcut|sort:date/sonarqube/4nJ-8VYOTuo/Tb1AnQC2AAAJ so it would be nice to have the same behavior for other rules affected.

Thanks and cheers,
Sandro

[nixel2007 (Nikita Fedkin)]

Maybe this should be transfered to “bug/false-positve” category, so sonar-java maintainers could have a look?

[Nicolas_Harraudeau]

Hi @sandro-h and sorry for the very long delay,

I can reproduce the problem on SonarQube 8.6.0 for both rules. I’ll create tickets to fix these False Positives.

Cheers,

[Nicolas_Harraudeau]

Here are the tickets: SONARJAVA-3668, SONARJAVA-3669.

Thank you for reporting these false positives!

[Nicolas_Harraudeau]

Hi @sandro-h @nixel2007.

I have an update on these two tickets:

• SONARJAVA-3669 will be fixed soon with a different approach. S2325 will not raise on empty methods.
• SONARJAVA-3668 on the other hand won’t be fixed. After reviewing some AspectJ projects we realized that some developers comment these methods to explain why they are empty. Thus the rule still seems to make sense for them.

Let me know if you have any question or remark.