CallingJdoPreclear tries to add the same instance to multiple 1-M relations

• Key: JDO-414
• Type: Bug
• Status: Resolved
• Resolution: Fixed
• Priority: Minor
• Assignee: Michelle Caisse
• Reporter: Marc Prud'hommeaux
• Component/s: tck2
• Affects Version/s: JDO 2 final
• Fix Version/s: JDO 2 maintenance release 1

Description

The CallingJdoPreclear test tries to add the same instance to multiple one-to-many relations in different instances, which may result in validation errors at commit time. The code that does this is:

        primaryObj.addChild(secondaryObj); // primaryObj contains one child; secondaryObj contains none. primaryObj is now dirty
        ...
        ternaryObj.addChild(secondaryObj);

Errors can be worked around by removing secondaryObj from primaryObj.children before adding it to ternaryObj. E.g.:

        primaryObj.addChild(secondaryObj); // primaryObj contains one child; secondaryObj contains none. primaryObj is now dirty
        primaryObj.children.remove(secondaryObj);

        ...
        ternaryObj.addChild(secondaryObj);

Alternately, the call to "primaryObj.addChild(secondaryObj);" can probably just be skipped ... it doesn't look like it would impact the test case.

Comments

[Michelle Caisse]

I agree that we should simple delete the call to "primaryObj.addChild(secondaryObj)". It doesn't appear to be necessary.

[Craig Russell]

If you skip the call to "primaryObj.addChild(secondaryObj)", you have to make primaryObj dirty in some other way, or you are not testing what the test case claims to be testing: new, clean, and dirty instance callbacks.

[Michelle Caisse]

Oops. What I meant was to skip the call to ternaryObj.addChild(secondaryObj) in the following code snippet:

        InstanceCallbackClass ternaryObj = new InstanceCallbackClass("ternaryObj", stillLaterDate, 3, 3.3, (short)-30, '3', null);
        pm.makePersistent(ternaryObj);
        ternaryObj.addChild(secondaryObj); // SKIP THIS LINE!!
        ternaryObj.addChild(primaryObj);
        t.commit();

[Craig Russell]

Ok, sounds good.

Fixed with revision: 475051