Bugzilla – Full Text Bug Listing
| Summary: | [1.5] VerifyError: Expecting to find object/array on stack | ||
|---|---|---|---|
| Product: | [Eclipse Project] JDT | Reporter: | Johan Compagner <jcompagner> |
| Component: | Core | Assignee: | Philipe Mulet <philippe_mulet> |
| Status: | VERIFIED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | ||
| Version: | 3.1 | ||
| Target Milestone: | 3.1 M5 | ||
| Hardware: | PC | ||
| OS: | Windows XP | ||
| Whiteboard: | |||
this does work: d /= 100; instead of d = d/100; Reproduced We should indeed box the result of #doubleValue()->double into Double. I am wondering where the #doubleValue()->Double is defined; javac uses it but it isn't defined in Number/Double. Oops, ignore my previous comment. We simply misgenerate the boxing conversion. Added AutoboxingTest#test092. Problem comes from optimization for converting d=d/100 into d/=100, passing along wrong implicitConversion value; and thus losing need for boxing conversion. Fixed Verified in I20050214 |
In my build of eclipse: I20050201-0800 i have this test file: public class Test { public Test() { super(); } public Object convert(Object value) { Double d = (Double)value; d = (d/100); return d; } public static void main(String[] args) { Test test = new Test(); Object value = test.convert(new Double(50)); } } eclipse generates the wrong byte code for that: 5: aload_2 6: invokevirtual #21; //Method java/lang/Double.doubleValue:()D 9: ldc2_w #22; //double 100.0d 12: ddiv 13: invokevirtual #21; //Method java/lang/Double.doubleValue:()D 16: astore_2 javac: 5: aload_2 6: invokevirtual #3; //Method java/lang/Double.doubleValue:()D 9: ldc2_w #4; //double 100.0d 12: ddiv 13: invokestatic #6; //Method java/lang/Double.valueOf:(D)Ljava/lang/Double; 16: astore_2 and that is logical because the double must be converted to a Double again and that shouldn't be done with a doubleValue() method (that a double doesn't have)