Description welljim 2010-09-23 15:25:57 EDT

Build Identifier: 20100218-1602 (Glassfish v3.0, upgraded with EclipseLink 2.1)

Assume 2 tables that are comprised of compound Primary Keys. Also, part of the
PK of table 2 is a FK of table 1. The following example illustrates this case:

DEPARTMENT
----------
ID -> (PK)
DEVISION_ID -> (PK)


PROJECT
--------
DEPT_ID -> (PK, FK)
DEVISION_ID -> (PK, FK)
NAME -> (PK)



Given that part of Project's PK is also a FK to Department (i.e., a project can
only be created under an existing department), this is mapped like so:

@Entity 
@IdClass(ProjectId.class)
public class Project {
    @Id
    @ManyToOne
    private Department dept;

    @Id
    private String name;

    // ...
}

public class ProjectId implements Serializable {
    private String name;
    private DeptId dept;

    public ProjectId() {}
    // ...
}

@Entity 
@IdClass(DeptId.class)
public class Department {
    @Id
    private int id;

    @Id
    private int devisionId;


    @OneToMany(mappedBy = "dept")
    private List<Project> projects;

    // ...
}

public class DeptId implements Serializable {
    private int id;
    private int devisionId;

    public DeptId() {}
    // ...
}



This works correctly when run in Glassfish v3.0 with EclipseLink 2.1, however
the Eclipse editor insists on underlining with red the annotation "@id" of
field "dept" in class "Project". Hovering the mouse over the underline brings
up the message "Column 'dept' cannot be resolved". In class Department, the
editor underlines the mappedBy name ("dept") with the message "Attribute named
'dept' has invalid mapping for this relationship".



Reproducible: Always

Steps to Reproduce:
1. Recreate the tables and entities described above
2. Notice that the Eclipse editor underlines the @Id and mappedBy as incorrect
3. Populate the table with some random data and run the configuration in
Glassfish or other compatible engine
4. The configuration runs as expected, and all find(), merge(), remove()
operations work without a problem