There is this JShell bug: https://bugs.openjdk.java.net/browse/JDK-8129337

Which basically comes down to getting two errors from an undefined symbol in this case:

    /tmp 1150>> cat > T.java
    class T {
      void f() { if (tttt) System.out.println("T"); }
    }
    /tmp 1151>> javac T.java
    T.java:2: error: cannot find symbol
      void f() { if (tttt) System.out.println("T"); }
                     ^
      symbol: variable tttt
      location: class T
    T.java:2: error: illegal start of type
      void f() { if (tttt) System.out.println("T"); }
                    ^
    2 errors

Is this the desired compiler behavior, or a chained error that should be cut-off?

Thanks,
Robert

 I think this is a javac bug - most likely the result of 'cannot resolve symbol' is an erroneous _class_ type, which means then javac complains because it's finding a type where an expression was required; obviously the logic that does the latter check is not skipping erroneous types appropriately.

Maurizio