Instance qualifier for static access warning and anonymous classes

[Tagir Valeev]

We in JetBrains, decided to remove a warning in this case, as there's
no obvious fix and it's not a suspicious or buggy code, it's likely to
be intended in a way it is.

With best regards,
Tagir Valeev.

On Fri, Apr 29, 2022 at 5:25 PM Maurizio Cimadamore
<maurizio.cimadamore at oracle.com> wrote:
>
> Well spotted - this seems like a dark corner.
>
> IMHO there are two possible interpretation, which depend on where we
> want to go with the warning:
>
> 1. static methods accessed through instance qualifiers are Bad, and we
> should try to wean developer off them over time (and align this with
> interface static methods). In this case the warning should stay (but of
> course the message needs to be fixed :-) ).
> 2. static methods accessed through instance qualifiers are a wart, but
> something we should be prepared to live with forever - best thing we can
> do in this space is to issue some "style" warnings, but nothing more
> will ever happen. In this case, for this example we should turn off the
> warning altogether.
>
> Maurizio
>
> On 29/04/2022 15:54, Tagir Valeev wrote:
> > Hello!
> >
> > Consider the following code:
> >
> > public class Demo {
> >      public static void main(String[] args) {
> >          var obj = new Object() {
> >              static void foo() {
> >                  System.out.println("Static method of anonymous class");
> >              }
> >          };
> >          obj.foo(); // Cannot replace instance qualifier with class reference
> >      }
> > }
> >
> > With -Xlint:static, javac reports the following warning:
> >
> > Demo.java:8: warning: [static] static method should be qualified by
> > type name, <anonymous Demo$1>, instead of by an expression
> >          obj.foo(); // Cannot replace instance qualifier with class reference
> >             ^
> > 1 warning
> >
> > It's questionable whether this warning should be reported at all, as
> > it's impossible to call the static method of anonymous class with
> > class name qualifier. But in any case, suggesting to use `<anonymous
> > Demo$1>` as a qualifier is invalid.
> >
> > With best regards,
> > Tagir Valeev.