JavaEE stub not working for Resource Leak Checker

[msridhar]

Input:

class Test {
    static void foo(javax.servlet.ServletResponse s) throws java.io.IOException {
	s.getWriter();
    }
}

In JavaEE.astub we have marked the getWriter() method as @NotOwning:

https://github.com/typetools/checker-framework/blob/master/checker/src/main/java/org/checkerframework/checker/mustcall/JavaEE.astub#L17-L17

And yet, compiling the above gives an error with 3.28.0:

$ ./checker-framework-3.28.0/checker/bin/javac -processor resourceleak -cp /path/to/javax.servlet-api-3.1.0.jar:. Test.java
Test.java:4: error: [required.method.not.called] @MustCall method close may not have been invoked on temp-var-1 or any of its aliases.
	s.getWriter();
	           ^
  The type of object is: java.io.PrintWriter.
  Reason for going out of scope: regular method exit
1 error

I used the following javax.servlet-api-3.1.0.jar in the classpath:

https://repo1.maven.org/maven2/javax/servlet/javax.servlet-api/3.1.0/javax.servlet-api-3.1.0.jar

[kelloggm]

There is a serious problem in AnnotatedTypeFactory#getDeclAnnotation, which (when called from MustCallConsistencyAnalyzer) is not producing the @NotOwning annotation from the stub file, contrary to its specification (checked via print-line debugging). I don't know what the cause of that problem is, but we should investigate further.

In the mean time, I'll open a PR that works around this problem in the short term by changing the @NotOwning annotations to @MustCall({}), delegating the work to the Must Call Checker.

[msridhar]

There is a serious problem in AnnotatedTypeFactory#getDeclAnnotation, which (when called from MustCallConsistencyAnalyzer) is not producing the @NotOwning annotation from the stub file, contrary to its specification (checked via print-line debugging). I don't know what the cause of that problem is, but we should investigate further.

In the mean time, I'll open a PR that works around this problem in the short term by changing the @NotOwning annotations to @MustCall({}), delegating the work to the Must Call Checker.

Is there any chance we could fix the AnnotatedTypeFactory#getDeclAnnotation method instead of doing a workaround? My concern is that I'm not really sure where else the RLC is currently relying on reading annotations from stub files and it is not working. It's possible you were planning to do that and the workaround is just a super-fast temporary fix.

[kelloggm]

It's possible you were planning to do that and the workaround is just a super-fast temporary fix.

That's correct, but I didn't have time to look into that this morning. It might also require expertise from either Mike or Suzanne, too, because it's probably a framework level issue. I opened the workaround PR to unblock you.

…

On Tue, Dec 27, 2022, 12:11 PM Manu Sridharan ***@***.***> wrote: There is a serious problem in AnnotatedTypeFactory#getDeclAnnotation, which (when called from MustCallConsistencyAnalyzer) is not producing the @NotOwning annotation from the stub file, contrary to its specification (checked via print-line debugging). I don't know what the cause of that problem is, but we should investigate further. In the mean time, I'll open a PR that works around this problem in the short term by changing the @NotOwning annotations to @MustCall({}), delegating the work to the Must Call Checker. Is there any chance we could fix the AnnotatedTypeFactory#getDeclAnnotation method instead of doing a workaround? My concern is that I'm not really sure where else the RLC is currently relying on reading annotations from stub files and it is not working. It's possible you were planning to do that and the workaround is just a super-fast temporary fix. — Reply to this email directly, view it on GitHub <#5472 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA6G5NQA6NGNXDCG2YNIUF3WPMPL5ANCNFSM6AAAAAATKARJOQ> . You are receiving this because you were assigned.Message ID: ***@***.***>