https://issues.apache.org/jira/browse/CAY-1507

I attached an Eclipse project illustrating the bug, too.  Also, as I
noted in the bug description, I had to have two relationships off the
master User record (Avatar and Role) in order to trip the bug.  When I
scaled down to just have User/Avatar, the bug went away.

I'll try to look into it more, but another set of eyes would
definitely be welcome.

Thanks,

mrg


On Mon, Nov 8, 2010 at 8:44 AM, Michael Gentry <[email protected]> wrote:
> If I turn optimistic locking off, it works as expected:
>
> INFO: UPDATE Users SET name = ? WHERE id = ?
> INFO: [bind: 1->name:'The Administrator', 2->id:200]
>
> I did a little digging on Friday and it looks like the snapshot
> doesn't contain the ID after the first commit which is what is causing
> the "WHERE id IS NULL" to occur on the second commit (I think).
>
> mrg
>
>
> On Sun, Nov 7, 2010 at 7:21 PM, Andrus Adamchik <[email protected]> 
> wrote:
>> Very odd. Please file a Jira. Maybe we can reproduce it with our unit tests? 
>> So what is optimistically locked in this case?
>>
>> Andrus
>

Reply via email to