RFR: 8261785: Calling "main" method in anonymous nested class crashes the JVM [v3]

[Henry Jen]

On Wed, 17 Mar 2021 08:55:54 GMT, Alan Bateman <alanb at openjdk.org> wrote:

> > > Using an anonymous class for the main class looks strange and hard to believe anyone is relying on this. I wonder if we should do more checking LauncherHelper.validateMainClass to reject cases like this.
> > 
> > 
> > I raised that same question, and people tends to agree launcher could reject anonymous/local classes. But pointed out that should require a CSR review. Therefore I chose to fix crash first, and we can file another ticket to address main class requirements.
> 
> The requirement for a CSR and release note should not be a concern here. I don't object to the fix but I think it would be very useful to document the main class and whether local or anonymous classes can be used, its accessibility, and the accessibility of the main method. We had to do something similar recently with the premain method (java.lang.instrument).

Absolutely we need to clarify that, however, the discussion of what should or should not be allowed may take some time. For example, if we limit such usage in launcher, should it be possible for custom launcher to start such a main method? Thus the recommendation to have a separate ticket for that.

The current java document mostly only say to load the specify the class name, however there is a sentence `By default, the first argument that isn't an option of the java command is the fully qualified name of the class to be called. If -jar is specified, then its argument is the name of the JAR file containing class and resource files for the application. The startup class must be indicated by the Main-Class manifest header in its manifest file.`. Not that it says `fully qualified name of the class`.

Filed [JDK-8263735](https://bugs.openjdk.java.net/browse/JDK-8263735) for the follow up, in the mean time, we should get this crash resolved.

-------------

PR: https://git.openjdk.java.net/jdk/pull/2999