Where and why JVM checks that the return type of e

问题:

I will try to answer both, please correct me if I am wrong:

Where: If a static method is being called using Classname.method() or using reflection then it doesn’t matter even if you change the return type of the calling method, the same method will still be called.

So JVM probably checks this in one of the native methods of jvm.cpp

methodHandle m (THREAD, init_klass->find_method(vmSymbols::object_initializer_name(),> vmSymbols::void_method_signature()));

if (m.is_null()) { ------ THROW_MSG_0 ………..

Why: Although its useless to return a value from main, as java does not do anything with it but if we try to change the return type of main to int for example, JVM throws

public static int main(String[] args) { return 1;
}

java.lang.NoSuchMethodError: main Exception in thread "main"

So may be Java mandates the use of same signature for entry method main() to maintain a symmetry in all Java programs written.

回答1:

From what I can gather, the reason main returns void in Java is threads.

C and C++ were both designed before multithreading was a common idiom, while threads were an integral part of Java from its conception. In any kind of non-trivial (multi-threaded) program, there is more than one thread, and so in reality your program never runs linearly from start to end of main.

Since the JVM doesn't halt execution until all non-daemon threads have finished running, returning from the main method doesn't mean your program ended.

With that in mind, void indeed seems like the most suited return type for main.

回答2:

As to "why":

I remember in the olden days on the Mac (OS 7 or so), the Mac JVM would accept a static void main() without any args (because the Mac had no command-line). That is gone now.

I suppose strict and unambigious behaviour is beneficial. Otherwise you'd end up with programs that work on some platforms and not on others for rather silly reasons. As you point out, any return value from main is discarded anyway.