C++ broker support for message priority, message TTL and bouncing unroutable messages

[Aidan Skinner <ai...@apache.org>]

Hey,

I've been converting a few test cases in systests that aren't using
QpidTestCase, and had to add ReturnUnroutableMandatoryMessageTest,
PriorityTest and TimeToLiveTest to the 010ExcludeList. Are those
features implemented in the C++ broker?

- Aidan
-- 
Apache Qpid - World Domination through Advanced Message Queueing
http://qpid.apache.org

Re: C++ broker support for message priority, message TTL and bouncing unroutable messages

[Martin Ritchie <ri...@apache.org>]

2009/1/14 Aidan Skinner <ai...@gmail.com>:
> On Wed, Jan 14, 2009 at 1:59 PM, Robert Godfrey <ro...@gmail.com> wrote:
>
>> ReturnUnroutableMandatoryMessageTest should be moved into some sort of
>> protocol version specific test folder.  For instance that test
>> *should* pass if run against Rabbit or OpenAMQ (do we split out or
>> system tests between generic AMQP and "Qpid specific"?, but the notion
>> of "mandatory" is different in 0-10 and will be different again in
>> 1-0...
>
> We don't do any of that sort of splitting. We probably should, fixing
> up systests is a bit of an open ended task though. ;)
>
> - Aidan
>
> --
> Apache Qpid - World Domination through Advanced Message Queueing
> http://qpid.apache.org

I'd like to see some thought and design put in to our testing so we
can have the tests:

- Specify or Create configuration that is given to the broker vm/java/c++
- Specify the protocol that the test should be run or excluded from
- Specify the broker type that the test should be run or excluded from

My thoughts on this would be to use some sort of annotation that way
you can tell when editing a test what it will run against rather than
having to check a bunch of files and build.xml files for exclude
details.

Thoughts

Martin

-- 
Martin Ritchie

Re: C++ broker support for message priority, message TTL and bouncing unroutable messages

[Aidan Skinner <ai...@gmail.com>]

On Wed, Jan 14, 2009 at 1:59 PM, Robert Godfrey <ro...@gmail.com> wrote:

> ReturnUnroutableMandatoryMessageTest should be moved into some sort of
> protocol version specific test folder.  For instance that test
> *should* pass if run against Rabbit or OpenAMQ (do we split out or
> system tests between generic AMQP and "Qpid specific"?, but the notion
> of "mandatory" is different in 0-10 and will be different again in
> 1-0...

We don't do any of that sort of splitting. We probably should, fixing
up systests is a bit of an open ended task though. ;)

- Aidan

-- 
Apache Qpid - World Domination through Advanced Message Queueing
http://qpid.apache.org

Re: C++ broker support for message priority, message TTL and bouncing unroutable messages

[Robert Godfrey <ro...@gmail.com>]

2009/1/14 Gordon Sim <gs...@redhat.com>:
> Aidan Skinner wrote:
>>
>> Hey,
>>
>> I've been converting a few test cases in systests that aren't using
>> QpidTestCase, and had to add ReturnUnroutableMandatoryMessageTest,
>> PriorityTest and TimeToLiveTest to the 010ExcludeList. Are those
>> features implemented in the C++ broker?
>
> Ttl is implemented on the broker; message priorities are not. The mandatory
> flag has changed somewhat in 0-10 so that test would likely need to be
> adapted (the c++ broker doesn't yet issue a reject if the accept-mode is set
> to explicit).
>

ReturnUnroutableMandatoryMessageTest should be moved into some sort of
protocol version specific test folder.  For instance that test
*should* pass if run against Rabbit or OpenAMQ (do we split out or
system tests between generic AMQP and "Qpid specific"?, but the notion
of "mandatory" is different in 0-10 and will be different again in
1-0...

-- Rob

Re: C++ broker support for message priority, message TTL and bouncing unroutable messages

[Aidan Skinner <ai...@gmail.com>]

On Wed, Jan 14, 2009 at 1:26 PM, Gordon Sim <gs...@redhat.com> wrote:

> Ttl is implemented on the broker; message priorities are not. The mandatory
> flag has changed somewhat in 0-10 so that test would likely need to be
> adapted (the c++ broker doesn't yet issue a reject if the accept-mode is set
> to explicit).

That all makes sense. I've filed QPID-1575 with the log files for the
TimeToLiveTest failures, which sounds like it should work.

- Aidan

-- 
Apache Qpid - World Domination through Advanced Message Queueing
http://qpid.apache.org

Re: C++ broker support for message priority, message TTL and bouncing unroutable messages

[Gordon Sim <gs...@redhat.com>]

Aidan Skinner wrote:
> Hey,
> 
> I've been converting a few test cases in systests that aren't using
> QpidTestCase, and had to add ReturnUnroutableMandatoryMessageTest,
> PriorityTest and TimeToLiveTest to the 010ExcludeList. Are those
> features implemented in the C++ broker?

Ttl is implemented on the broker; message priorities are not. The 
mandatory flag has changed somewhat in 0-10 so that test would likely 
need to be adapted (the c++ broker doesn't yet issue a reject if the 
accept-mode is set to explicit).