Essentially this results in creation of a few HashMaps... Not very
expensive by itself. My note on performance was mostly related to the
fact that DataContext level caching will not have a chance to play
significant role in a short-lived context.
Andrus
On Mar 18, 2008, at 2:35 AM, Gary Jarrel wrote:
> Thank you for the quick reply.
>
> I know I should probably do some of my own profiling, but is the
> process of
> creating a new datacontext expensive computation wise?
>
> Thank you!
>
> Gary
>
> On Tue, Mar 18, 2008 at 4:55 AM, Andrus Adamchik <andru..bjectstyle.org
> >
> wrote:
>
>> The simplest thing to do is to limit the scope of the thread
>> DataContext to the 'onMessage' method. I.e. create a new context on
>> entry, bind it to the current thread, and unbind it in the "finally"
>> block... The performance impact is of course that you need to refetch
>> everything related to your processing on every call, which may or may
>> not be a problem.
>>
>> Andrus
>>
>>
>> On Mar 17, 2008, at 4:18 PM, Gary Jarrel wrote:
>>
>>> Hi All!
>>>
>>> One of the requirements of the current project I'm working on is to
>>> use a
>>> j2ee environment and message driven beans in an application server
>>> such as
>>> glassfish.
>>>
>>> There is also a requirement to use a custom DAO library which relies
>>> on a
>>> thread bound data context.
>>>
>>> Basically the message bean receives a text message via JMS an
>>> prior to
>>> processing it, it must be stored in the database, then, once
>>> processed the
>>> results are also stored in a database.
>>>
>>> I've always tried to avoid EJBs however, in this case, my question
>>> is: has
>>> anyone got any "best practice" advice on how a data context should
>>> be
>>> created and bound to the thread in a message bean environment.
>>>
>>> The message beans are quite simple, in that they only receive the
>>> messages,
>>> an then call upon spring POJOs to do the processing of the messages.
>>>
>>> Thank you in advance.
>>>
>>> garyj
>>
>>
This archive was generated by hypermail 2.0.0 : Tue Mar 18 2008 - 04:09:15 EDT