Re: [eclipselink-users] Possible Bug: @ReturnUpdate drops timestamps fr

[Tim Hollosy <hollosyt@xxxxxxxxx>]

I turned off the Shared Cache - and now it's working perfectly, even
with @ReturnUpdate. I'd still like to use a shared cache however, but
I just don't have any time to mess around with this at the moment. Any
other things I could try to narrow this down?

./tch



On Fri, Jun 25, 2010 at 1:08 PM, Tim Hollosy <hollosyt@xxxxxxxxx> wrote:
> Ok, so I now see my refresh is pulling out the correct data for
> modified on. But subsequent find calls still are using the stale
> modified on date!
>
> I added back on my @ReturnUpdate, again I see the correct data in the
> entity when I do a refresh, however on subsequent finds it's pulling
> out the stale modified_on data!
>
> My MappedSuperclass is annotated with @Cache(refreshOnlyIfNewer=false)
> if that matters.
>
> This is very bizarre!
>
> ./tch
>
>
>
> On Fri, Jun 25, 2010 at 12:46 PM, Tim Hollosy <hollosyt@xxxxxxxxx> wrote:
>> I'm having very odd behavior with modified on's now. It was working
>> when I commented out my @ReturnUpdate, now even that stopped working.
>> Now it's not refreshing at all!
>>
>> I added the system property to my server start, didn't matter.
>>
>> I'm using static weaving, and my modifed_on column is part of a
>> @MappedSuperclass. In debug mode, i watch the refresh, and all the
>> columns are highlighted like they were changed EXCEPT the columns in
>> my @MappedSuperclass, so I suspect that is the actual culprit here.
>>
>> Ideas?
>>
>> ./tch
>>
>>
>>
>> On Fri, Jun 25, 2010 at 11:58 AM, Andrei Ilitchev
>> <andrei.ilitchev@xxxxxxxxxx> wrote:
>>> Also does the same happen with other type (String)?
>>>
>>> Could you please try running without change tracking
>>> (-Declipselink.weaving.changetracking=false).
>>> It the problem still there?
>>>
>>> On 25/06/2010 11:21 AM, Tim Hollosy wrote:
>>>>
>>>> This is very odd,
>>>> When using @ReturnUpdate to manage a Modified_On column (Oracle Date
>>>> type). The timestamp is missing from the refreshed object until the
>>>> container is restarted.
>>>>
>>>> It's really odd, I do a save and then a refresh in the refreshed
>>>> object, and in all future ReadAllQueries even against that object the
>>>> timestamp is coming back as midnight. This happens on
>>>> java.sql.Timestamp types as well as java.util.Date with @Temporal and
>>>> a timestamp.
>>>>
>>>> It's funny because it doesn't happen with @ReturnInsert any clue at
>>>> all, this seems to weird to be an actual bug and I'm not going to rule
>>>> out user error. Any comments before I file a bug report?
>>>>
>>>>
>>>> ./tch
>>>> _______________________________________________
>>>> eclipselink-users mailing list
>>>> eclipselink-users@xxxxxxxxxxx
>>>> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
>>>>
>>>
>>> _______________________________________________
>>> eclipselink-users mailing list
>>> eclipselink-users@xxxxxxxxxxx
>>> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
>>>
>>
>