Restrict subtraction zero peepholes to integer operandspr254-fixes

Require resolved integer mode in chapter 25 so zero subtraction preserves floating-point signed zero. Keep nonnumeric operands in the graph for subtraction diagnostics.

Add signed-zero coverage from chapter 12 and null-subtraction diagnostics from chapter 22. Confirmed the regressions fail on the original PR. Full Make suites pass in chapters 4-25 (5462 tests), including fresh chapter 25 sys.o compilation.

1 files changed, 7 insertions, 0 deletions

diff --git a/chapter17/src/test/java/com/seaofnodes/simple/Chapter12Test.java b/chapter17/src/test/java/com/seaofnodes/simple/Chapter12Test.java
index e81f9845..f7efde96 100644
--- a/chapter17/src/test/java/com/seaofnodes/simple/Chapter12Test.java
+++ b/chapter17/src/test/java/com/seaofnodes/simple/Chapter12Test.java
@@ -8,6 +8,13 @@ import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.fail;
 
 public class Chapter12Test {
+    @Test public void testSubZeroFloat() {
+        var stop = new Parser("flt x = arg; return 0-x;").parse().iterate();
+        // Compare strings so +0.0 and -0.0 remain distinct.
+        assertEquals("0.0", Evaluator.evaluate(stop,0).toString());
+        assertEquals("-1.0", Evaluator.evaluate(stop,1).toString());
+    }
+
     @Test public void testDeadNumericReturns() {
         for( String[] test : new String[][] {
             {"if(1) return 7; else return 2.5;", "7"},