[
https://issues.apache.org/jira/browse/OPENJPA-504?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Christiaan resolved OPENJPA-504.
--------------------------------
Resolution: Fixed
Fix Version/s: 1.2.0
1.1.0
Fixed due to issue OPENJPA-567
> Duplicate order numbers in relationshiptable for the same List field
> --------------------------------------------------------------------
>
> Key: OPENJPA-504
> URL: https://issues.apache.org/jira/browse/OPENJPA-504
> Project: OpenJPA
> Issue Type: Bug
> Environment: Kodo 4.1.4
> Reporter: Christiaan
> Fix For: 1.1.0, 1.2.0
>
>
> I now and then encounter entries in the relationship tables for List fields
> (id, element, ordr) non-unique entries for id, ordr. (Two elements are placed
> on the same index in the same list). I am not sure how to reproduce this, but
> I do have a question regarding the code in
> StoreCollectionFieldStrategy.load():
> ChangeTracker ct = null;
> if (field.getTypeCode() == JavaTypes.ARRAY)
> coll = new ArrayList();
> else {
> coll = sm.newProxy(field.getIndex());
> if (coll instanceof Proxy)
> ct = ((Proxy) coll).getChangeTracker();
> }
> // load values
> Result res = union.execute(store, fetch);
> try {
> int seq = 0;
> while (res.next()) {
> if (ct != null && field.getOrderColumn() != null)
> seq = res.getInt(field.getOrderColumn());
> add(store, coll, loadElement(sm, store, fetch, res,
> resJoins[res.indexOf()]));
> }
> if (ct != null && field.getOrderColumn() != null)
> ct.setNextSequence(seq + 1);
> } finally {
> res.close();
> }
> If I am not mistaken, this initializes the changetracker for list fields for
> persistent objects loaded from the database with an empty list attribute
> (res.next() = false) with order number of "1", whereas objects with non-empty
> List fields which are persisted start with an order number of "0". Could this
> be the reason that I now and then encounter duplicate entries in the
> relationship table with ordernr "1"? Actually this is also the reason why
> encounter duplicates for other numbers, since order number is zero based
> (next sequence number is initialized to size()), but as described, some lists
> start counting from "1".
> kind regards,
> Christiaan
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
