Hi Bryan,
ObjectStore.retainedSnapshotMap should keep snapshots of all modified
uncommitted objects. If this is not the case, I'd really like to track
down the reason. Doing "getSnapshot()" instead of
"getRetainedSnapshot()" should actually be OK as a temp workaround, but
not as a permanent fix, as cross-context synchronization consistency
will be affected.
Could you please open a bug report with your findings? If you have a
code example to reproduce the problem, please attach it to the report -
this would be awesome.
Thanks
Andrus
On Oct 14, 2004, at 12:16 PM, Bryan Lewis wrote:
> I'm getting a NullPointerException in version 1.1-RC1.
>
> org.objectstyle.cayenne.access.ContextCommit.appendOptimisticLockingAtt
> ributes(ContextCommit.java:564)
> org.objectstyle.cayenne.access.ContextCommit.prepareUpdateQueries(Conte
> xtCommit.java:426)
> org.objectstyle.cayenne.access.ContextCommit.commit(ContextCommit.java:
> 156)
> org.objectstyle.cayenne.access.DataContext.commitChanges(DataContext.ja
> va:1231)
> org.objectstyle.cayenne.access.DataContext.commitChanges(DataContext.ja
> va:1201)
>
> I've stepped through the code; the snapshot being retrieved at line
> 556 is null, causing the exception a few lines later in
> snapshot.get(name).
>
> The DataObject is a normal object with state = modified; a new object
> is being added to its to-many relationship. There's only one
> DataContext in use. If I tell the Modeler not to use optimistic
> locking for the modified object, the exception doesn't happen because
> appendOptimisticLockingAttributes() isn't called.
>
> I copied the snapshot-getting line 556 to my own code for an
> experiment.
> If I replace getRetainedSnapshot() with getSnapshot(objectId,
> dataContext), a non-null snapshot is returned. That might not be the
> right solution... I just wanted to confirm that the null result was
> coming from getRetainedSnapshot().
This archive was generated by hypermail 2.0.0 : Thu Oct 14 2004 - 20:06:38 EDT