Bug 251523

Summary:

[1.5][compiler] Should still flag Enum#valueOf override

Product:

[Eclipse Project] JDT

Reporter:

Philipe Mulet <philippe_mulet>

Component:

Core

Assignee:

Kent Johnson <kent_johnson>

Status:

VERIFIED FIXED

QA Contact:

Severity:

normal

Priority:

CC:

david_audel, frederic_fusier

Version:

3.5

Target Milestone:

3.5 M3

Hardware:

OS:

Windows XP

Whiteboard:

Attachments:

Description	Flags
Proposed patch	none
In progress patch	none

Description Philipe Mulet

2008-10-21 08:05:39 EDT

Integration 10/21/2008.

Regression introduced by fix for bug 146768
The following code should be rejected:

public enum X {
	;
	private X valueOf(String arg0) { return null; }
}

Comment 1 Philipe Mulet

2008-10-21 08:25:05 EDT

Created attachment 115681 [details]
Proposed patch

Comment 2 Philipe Mulet

2008-10-21 09:31:07 EDT

The patch is not good enough.

BTW, we also need to handle properly a more complex case:

public enum X {
        ;
        private X valueOf(String arg0) { return null; }
        private X valueOf(String arg0) { return null; }
        private X valueOf(String arg0) { return null; }
        
        void foo() {
        	valueOf("");
        }
} 

The generated problem type should still dump one valueOf method for consistency between batch/incremental compile. 

Ideally, it should keep the implicit synthetic method (considering the message indicates it taking precedence).

Comment 3 Philipe Mulet

2008-10-21 09:42:16 EDT

Created attachment 115694 [details]
In progress patch

This patch still doesn't address the problem Type issue. It does ensure reporting a problem against each dup methods (including the first one), and should handle enum specials ok (nuking all user dup methods).

Code in HEAD seems to only report a problem against subsequent dup methods, and not original one any longer (where I noticed some tests expected all methods to be flagged)

Comment 4 Kent Johnson

2008-10-21 14:00:21 EDT

Released enum test164 for 3.5M3

Was fixed by new patch for bug 146760

Comment 5 David Audel

2008-10-28 10:11:54 EDT

Verified for 3.5M3 using I20081026-2000 build.