On a slightly different note, but related. We have done some interop testing for the forthcoming M2 release of Qpid. This consisted of some very simple test cases between the Qpid C++, Java and .Net clients on the Qpid Java broker. This is really the only combination of AMPQ 'components' where there has been a concerted effort to verify interopability.
There is a specification for these tests at: http://cwiki.apache.org/qpid/interop-testing-specification.html It would be nice to start expanding on this, for example, running interop tests with the three clients already mentioned, over the Qpid C++ broker, and other implementations too. I doubt that the Rabit client will pass the tests without modification, because AFAIK our 0.8 clients had to fill in some missing pieces of the 0.8spec, specifically, defining type mappings for field tables, and a binding URL format for reply-to destinations. I'm assuming Rabit didn't copy these for their client? Rupert On 16/08/07, Robert Godfrey <[EMAIL PROTECTED]> wrote: > > Hi, at some point I added implementations of access ticket to the Qpid > client to allow us to interoperate with the RabbitMQ broker... if the code > has disappeared I can put it back in. > > At present Qpid does not implement the access ticket / realm parts of > AMQP. > > Hope this helps, > Rob > > On 16/08/07, Matthias Radestock <[EMAIL PROTECTED]> wrote: > > > > Tanmay, > > > > Goel, Tanmay wrote: > > > when I try to run a RabbitMQ client (attached, > > > SimpleProducer.java) against Qpid broker (with the default > configuration > > > provided when I first downloaded it, no change), I get the following > > > exception. The exception is thrown on this like of code: * > > > > > > *int ticket = ch.accessRequest("/test");* > > > > > > *Intuitively, this method is not provided by the Qpid broker. Is it?* > > > > You are probably right and this method is indeed unimplemented in Qpid. > > It certainly wasn't when we ran the tests ~6 months ago, as noted in > > http://www.rabbitmq.com/interoperability.html > > > > You can probably just skip that request and use a ticket number of 0, > > with RabbitMQ's lax_mode enabled (which it is by default - see above > > page). > > > > > *Also, there is no "client-name" passed as a parameter at the time of > > > establishing connection to the RabbitMQ connection constructor but is > > > passed to the Qpid connection constructor. If and how will this > effect?* > > > > I have no idea what the "client-name" value is for. Perhaps the QPid > > client passes that value as part of the peer properties exchanged during > > connection negotiation, in which case it's informational only, i.e. it > > shouldn't affect the behaviour. > > > > > *Alternatively, when I tried to run a Qpid client against a RabbitMQ > > > broker (again, with the default settings, virtual hosts, etc). I got > the > > > following error message at the first line of code: _connection = new > > > AMQConnection("localhost", 5672, "guest", "guest", "clientid", > > "/data");* > > > > What do the RabbitMQ server logs (stdout, rabbit.log, rabbit-sasl.log) > > say? > > > > What version of QPid are you running? The M1 release? > > > > Regards, > > > > Matthias. > > > > _______________________________________________ > > rabbitmq-discuss mailing list > > [EMAIL PROTECTED] > > http://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss > > >
