Bug 108263 - [1.5][compiler] Constants initilialization doesn't work inside Annotation
Summary: [1.5][compiler] Constants initilialization doesn't work inside Annotation
Status: VERIFIED FIXED
Alias: None
Product: JDT
Classification: Eclipse Project
Component: Core (show other bugs)
Version: 3.1   Edit
Hardware: PC Windows 2000
: P3 normal (vote)
Target Milestone: 3.1.1   Edit
Assignee: Philipe Mulet CLA
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-08-29 09:11 EDT by David Audel CLA
Modified: 2005-09-26 12:04 EDT (History)
2 users (show)

See Also:

Attachments
Proposed fix (8.48 KB, patch)
2005-08-29 17:00 EDT, Olivier Thomann CLA 		no flags Details | Diff
Regression test (2.38 KB, patch)
2005-08-29 17:00 EDT, Olivier Thomann CLA 		no flags Details | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description David Audel CLA 2005-08-29 09:11:05 EDT 
build I20050823-0800 + jdtcore head

1) create A.java

public @interface A {
  public static final Integer foo = B.zzz; 
  public static final int foo3 = B.zzz2; 
}
class B {
  public static final Integer zzz = new Integer(0);
  public static final int zzz2 = 0;
}
2) compile
jdtcore compiler found one error
"The annotation field A.foo must be initialized with a constant expression"

and javac none
Comment 1 Olivier Thomann CLA 2005-08-29 14:56:55 EDT 
If the annotation is converted to an interface, it works fine.
Comment 2 Olivier Thomann CLA 2005-08-29 16:10:26 EDT 
According to the specs, there is no need to have a constant expression as the
initialization of such a field.
The field simply needs to be initialized. The treatement should be the same as
the field for an interface.
Such a field without an initialization should report the error:
The blank final field ... may not have been initialized

and not the error :
The annotation field ... must be initialized with a constant expression

I will attach a patch that makes the handling of annotations close to the one of
interfaces.
Comment 3 Olivier Thomann CLA 2005-08-29 17:00:00 EDT 
Created attachment 26601 [details]
Proposed fix
Comment 4 Olivier Thomann CLA 2005-08-29 17:00:23 EDT 
Created attachment 26602 [details]
Regression test
Comment 5 Olivier Thomann CLA 2005-08-30 08:03:01 EDT 
All tests passed.
Comment 6 Olivier Thomann CLA 2005-08-30 10:23:13 EDT 
This could be a candidate for 3.1.1.
Comment 7 Philipe Mulet CLA 2005-09-06 05:28:11 EDT 
Why does the patch remove the check for annotation type constructor ? These are
still forbidden.
Comment 8 Philipe Mulet CLA 2005-09-06 05:40:30 EDT 
Ok, it actually moves it elsewhere, to match the interface handling. This is
good, though not mandated by this defect.
Comment 9 Philipe Mulet CLA 2005-09-06 07:07:28 EDT 
+1 for 3.1.1.

Patch looks mostly just fine. Will integrate and adjust a little.
Will also add test for recovery of local annotation type decl.
Comment 10 Philipe Mulet CLA 2005-09-06 07:09:08 EDT 
Kent - in BinaryTypeBinding constructor, why do we have only ?

	// TODO (kent) what about ANNOTATION_TYPE_DECL case ?
	if (binaryType.getKind() == IGenericType.INTERFACE_DECL)
		this.modifiers |= AccInterface;


Feels like annotation type case is not handled. But, do we need this statement
at all ? Modifiers should already contain the information I suppose.
Comment 11 Philipe Mulet CLA 2005-09-06 07:45:22 EDT 
Fixed. Added AnnotationTest#test177, AnnotationDietRecoveryTest#test0010.
Updated impacted existing tests.

(did not touch BinaryTypeBinding for now)
Comment 12 David Audel CLA 2005-09-21 10:44:44 EDT 
Verified in I20050921-0010 for 3.2M2
Comment 13 David Audel CLA 2005-09-26 12:04:47 EDT 
Verified using M20050923-1430 for 3.1.1