private native Throwable fillInStackTrace(int dummy);
This method is called with dummy=0
when an Exception is created. What is the meaning of dummy
? Is it the depth of the stack trace to construct?
Update:
public class MyEx extends RuntimeException{
@Override
public synchronized Throwable fillInStackTrace() {
Method method = null;
try {
Class<?>[] classArray = new Class<?>[1];
classArray[0] = int.class;
method =Throwable.class.getDeclaredMethod("fillInStackTrace", classArray);
method.setAccessible(true);
Object obg = method.invoke(this, 6);
StackTraceElement[] trace = ((MyEx) obg).getStackTrace();
System.out.println();
} catch (NoSuchMethodException e) {
e.printStackTrace();
} catch (IllegalAccessException e) {
e.printStackTrace();
} catch (InvocationTargetException e) {
e.printStackTrace();
}
return this;
}
}
It seems that dummy is really dummy, it doesn't matter what value I put in, the result will be the same...
I wanted to limit the stack trace size to 3 to consume less memory and also creating exceptions would be faster from an execution point of view. I have a real use case where I need many exceptions but with shallow stack traces.
It doesn't mean anything. It is a dummy argument.
The native code implementation completely ignores the argument. For example, in OpenJDK 11, the bridge method implementation is as follows:
/*
* Fill in the current stack trace in this exception. This is
* usually called automatically when the exception is created but it
* may also be called explicitly by the user. This routine returns
* `this' so you can write 'throw e.fillInStackTrace();'
*/
JNIEXPORT jobject JNICALL
Java_java_lang_Throwable_fillInStackTrace(JNIEnv *env,
jobject throwable, jint dummy)
{
JVM_FillInStackTrace(env, throwable);
return throwable;
}
As you can see, the dummy
argument is ignored.
If you are looking for a way to limit the depth of a stacktrace, the supported way to do this is to use the -XX:MaxJavaStackTraceDepth=depth
option.