Compiler could be smarter when dealing with exhaustive switch and final variables?

[David Alayachew]

Terribly sorry Jan, I have been using JDK 18.0.2 this entire time.

For JDK 19, everything works, as you said. However, for JDK 18.0.2, this is
my repeatable example and the corresponding error.

public class TestExhaustivenessBug
{

   sealed interface SomeInterface {}
   record First() implements SomeInterface {}
   record Second() implements SomeInterface {}

   private void exhaustiveIfWithFinalUninitialized()
   {

      SomeInterface abc = new First();

      final int numA;
      final int numB;

      if (abc instanceof First)
      {

         numA = -1;
         numB = -1;

      }

      else if (abc instanceof Second)
      {

         numA = -2;
         numB = -2;

      }

      else
      {

         throw new IllegalStateException("This shouldn't be possible!");

      }

      System.out.println(numA + "");
      System.out.println(numB + "");

   }

   private void exhaustiveSwitchWithFinalUninitialized()
   {

      final int numA;
      final int numB;

      SomeInterface abc = new First();

      switch (abc)
      {

         case First f  ->
         {

            numA = 1;
            numB = 0;

         }

         case Second s ->
         {

            numA = 0;
            numB = 1;

         }

      }

      System.out.println(numA + "");
      System.out.println(numB + "");

   }


}

And here is the error I received.

TestExhaustivenessBug.java:74: error: variable numA might not have been
initialized
      System.out.println(numA + "");
                         ^
TestExhaustivenessBug.java:75: error: variable numB might not have been
initialized
      System.out.println(numB + "");
                         ^
Note: TestExhaustivenessBug.java uses preview features of Java SE 18.
Note: Recompile with -Xlint:preview for details.
2 errors

Apologies for the trouble and confusion.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/amber-dev/attachments/20221008/a612274b/attachment.htm>