Bugzilla – Full Text Bug Listing
| Summary: | JACKS - The null literal should not be considered as a constant expression | ||
|---|---|---|---|
| Product: | [Eclipse Project] JDT | Reporter: | Olivier Thomann <Olivier_Thomann> |
| Component: | Core | Assignee: | Philipe Mulet <philippe_mulet> |
| Status: | VERIFIED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | ||
| Version: | 2.1 | ||
| Target Milestone: | 2.1 M4 | ||
| Hardware: | PC | ||
| OS: | Windows 2000 | ||
| Whiteboard: | |||
Another test case:
class A {
void foo() {
boolean b = "" + null == "null";
System.out.println(b);
}
}
The first example seems right to me. We are more optimized. Do you have an example showing an actual problem ? In the second example we don't do the string concatenation. We inline the constant. In fact we should never inline the constant when it is a null literal. I am not sure we "can" optimize the first case. Took String concatenation inlining when null is an argument. Keeping internal constant folding with null literals still. Fixed Removed internal null constants. Fixed Verified. |
Using 1112, we inline null == null. class A { void foo() { boolean b = null == null; System.out.println(b); } } Code attribute: /* Stack: 2, Locals: 2 */ 0 iconst_1 1 istore_1 2 getstatic #20 <Field java.lang.System#out java.io.PrintStream> 5 iload_1 6 invokevirtual #26 <Method java.io.PrintStream#println(boolean arg) void> 9 return We should have instead: Method void foo() 0 aconst_null 1 ifnonnull 8 4 iconst_1 5 goto 9 8 iconst_0 9 istore_1 10 getstatic #2 <Field java.io.PrintStream out> 13 iload_1 14 invokevirtual #3 <Method void println(boolean)> 17 return Javac 1.4 and jikes generate the right bytecode sequence.