Bugzilla – Full Text Bug Listing
| Summary: | static final char NegThree= (char)-3, -3 == NegThree returns true | ||
|---|---|---|---|
| Product: | [Eclipse Project] JDT | Reporter: | Jed Anderson <jed.anderson> |
| Component: | Core | Assignee: | Philipe Mulet <philippe_mulet> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | eclipse |
| Version: | 2.0 | ||
| Target Milestone: | 2.0 M6 | ||
| Hardware: | PC | ||
| OS: | Windows 2000 | ||
| Whiteboard: | |||
I think I find the problem.
As both operands are constant, the compiler optimizes the code, it computes the
result of the operation and just put this value in the byte code.
But I think there is a typo in the code which manages that, in
EqualExpression.computeConstant(TypeBinding, TypeBinding). With the following
modification, the behavior of the generated code is correct.
public final void computeConstant(TypeBinding leftTb, TypeBinding rightTb) {
if ((left.constant != NotAConstant) && (right.constant != NotAConstant)) {
constant =
Constant.computeConstantOperationEQUAL_EQUAL(
left.constant,
leftTb.id, <============= there is the modification
EQUAL_EQUAL,
right.constant,
rightTb.id);
if (((bits & OperatorMASK) >> OperatorSHIFT) == NOT_EQUAL)
constant = Constant.fromValue(!constant.booleanValue());
} else {
constant = NotAConstant;
}
}
Good find. Thanks. Fixed. |
Build: M5++ Compile the following with javac and run it with java (version 1.4.0-rc-b91) public class A { static final char NegThree= (char)-3; public static void main(String[] args) { System.out.println(-3 == NegThree); } } You will get false. Try this with Eclipse, and you will get true. This is wrong. It appears that the compiler is inlining the incorrect value, because if you remove the "final" modifier it runs correctly.