On 17 January 2013 21:51, Robbie Gemmell <[email protected]> wrote:

>
> On 17 January 2013 21:06, Hiram Chirino <[email protected]> wrote:
>
>> FYI there is a good reason to NOT preserve the message id.  It's fairly
>> common for clients to do things like:
>>
>> msg = session.createTextMessage();
>> msg.setText("Hello");
>> sender.send(msg);
>> msg.setText("World");
>> sender.send(msg);
>>
>> You don't want the 2nd msg sent /w the message id used in the 1st.
>>
>
> Agreed, this was the exact type of use case I was concerned with in an
> earlier message.
>
>
>> The only time your really want to do the id preservation is when a
>> 'foreign' jms message is being sent.  At least this is what ActiveMQ does.
>>  If a Qpid JMS message is sent on an ActiveMQ sender, the message id will
>> be preserved.
>>
>>
> How do you handle the situation above where the message was foreign, just
> let it duplicate the ID for the foreign message, given it it would be ever
> so slightly slightly less usual to send the foreign message twice?
>
>

(...err...obviously its not exactly the same situation, since you wouldnt
be able to call setText on the message given it had been recieved....but
the send twice bit still applies I guess :P )


>
>>
>> On Thu, Jan 17, 2013 at 1:07 PM, Rajith Attapattu <[email protected]
>> >wrote:
>>
>> > On Thu, Jan 17, 2013 at 12:43 PM, Robbie Gemmell
>> > <[email protected]> wrote:
>> > >>
>> > > Yep, 3272
>> >
>> > Gotcha.
>> >
>> > > Seperate as in store the provided String in a field which is separate
>> > from
>> > > the UUID in the header where we would normally store the passed in
>> > message
>> > > id value, and return this when getJMSMessageID is called.
>> >
>> > Thanks for clarifying. It makes sense.
>> >
>> >
>> > On a separate note : All though (based on the updated info) we don't
>> > need to consider preserving the original message-id for now, this may
>> > become important in the future.
>> > Consider a message being routed through a federated network, where it
>> > takes several hops before reaching the endpoint.
>> > One of those hops could very well be implemented using a JMS client.
>> > In this case the Qpid client will overwrite the message-id with a new
>> > one.
>> > The hops are merely forwarding/routing agents and in that sense
>> > tampering with the original message properties may not be a good idea.
>> > In such a situation it would be useful to have a configurable way to
>> > ensure the send-impl doesn't override the message id.
>> > I think we will be seeing such use cases in the near future.
>> >
>> > ---------------------------------------------------------------------
>> > To unsubscribe, e-mail: [email protected]
>> > For additional commands, e-mail: [email protected]
>> >
>> >
>>
>>
>> --
>>
>> **
>>
>> *Hiram Chirino*
>>
>> *Engineering | Red Hat, Inc.*
>>
>> *[email protected] <[email protected]> | fusesource.com | redhat.com*
>>
>> *skype: hiramchirino | twitter: @hiramchirino<
>> http://twitter.com/hiramchirino>
>> *
>>
>> *blog: Hiram Chirino's Bit Mojo <http://hiramchirino.com/blog/>*
>>
>
>

Reply via email to