> On Feb. 25, 2014, 11:17 a.m., Gordon Sim wrote:
> > Looks good!
> >
> > - BrokerAgent QMF not working with SWIG client - never receive response
> > messages.
> >
> > - Exception types not corretly propagated over AMQP 1.0 to swig client
> > e.g. raises LinkError instead of NotFound if queue not found
> >
> > - auto-delete appears not to work with 1.0,
> >
> > Can you be more specific? It should work, however there are some subtle
> > differences at present in how it is interpreted. For 1.0 it is currently
> > assumed to mean delete when not 'in use', meaning no sender or receiver
> > links. In 0-10 however it is specified to mean that the consumer count
> > drops to 0 from some larger value (assuming exclusive is not set, which
> > changes the meaning).
> >
> > - no TX support for 1.0.
> >
> > Indeed, known limitation at present.
> >
> > - Can't create an exclusive subscription over 1.0? This fails:
> > "excl_sub;{create:always, link:{x-subscribe:{exclusive:True}}}");
> >
> > By default receivers from an exchange will each create their own exclusive
> > subscription. You can have shared subscriptions by specifying the shared
> > capability for the node (and naming the link). However you are right that
> > there is no way to indicate exclusivity when consuming from a queue, and
> > that does seem like a gap worth addressing as it can be a valuable semantic
> > to enforce in some applications.
> >
>
> Gordon Sim wrote:
> Oops, forgot to respond to the first two...
>
> - BrokerAgent QMF not working with SWIG client - never receive response
> messages.
>
> I have successfully tested QMf over 1.0, however not using the
> qpidtoollibs. I suspect this may be due to it using the
> qpid.messaging.Message directly, which is different from the swigged one.
> Just a suspicion though.
>
> - Exception types not corretly propagated over AMQP 1.0 to swig client
> e.g. raises LinkError instead of NotFound if queue not found
>
> This looks like it is an issue with the broker and the c++ client, not
> specifically a swig issue. The original way of handling a non-existent node
> seems to have been changed inadvertently on the broker and the client doesn't
> handle that. I'll look into that further.
>
> Gordon Sim wrote:
> The exception issue should now be fixed. (For the record it wasn't a
> broker side change at all).
RE: - auto-delete appears not to work with 1.0, Can you be more specific?
All the auto delete tests fail because queues are not deleted as expected but
I haven't looked into it more than that. Will ping you when I have more detail
RE: I have successfully tested QMf over 1.0, however not using the
qpidtoollibs. I suspect this may be due to it using the qpid.messaging.Message
directly
Good point! I had that problem elsewhere but didn't think of it in regard to
qpidtoollibs. That shouldn't be hard to fix. Thanks!!
Re: The exception issue should now be fixed.
Great again, I'll re-instate specific catches and verify that it works.
- Alan
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/18387/#review35387
-----------------------------------------------------------
On Feb. 22, 2014, 12:49 a.m., Alan Conway wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/18387/
> -----------------------------------------------------------
>
> (Updated Feb. 22, 2014, 12:49 a.m.)
>
>
> Review request for qpid and Gordon Sim.
>
>
> Repository: qpid
>
>
> Description
> -------
>
> Summary of changes:
> - brokertest.py: configurable support for of swig vs. native and amqp0-10 vs.
> 1.0
> - default to swig+amqp1.0 if swig is available, native+amqp0-10 otherwise
> - reworked ha_tests to work with amqp1.0 and 0-10
> - bind, delete policies not supported by AMQP 1.0, switched to using
> BrokerAgent QMF.
> - catch general Exception rather than specific messaging exception types.
> - pass protocol:amqp1.0 connection-option to c++ test clients (qpid-send,
> qpid-receive)
> - misc. minor work-arounds
> - Swig python client improvements:
> - support for passing client_properties/properties.
> - expose AddressHelper pn_data read/write as PnData helper class
> - set sender/receiver capacity on creation
> - limited disposition support - rejected messages.
> - support for additional timeout parameters
> - expose messaging::Logger, allow log configuration to be set from python.
>
> Unresolved problems, marked with FIXME comments:
> - BrokerAgent QMF not working with SWIG client - never receive response
> messages.
> - Exception types not corretly propagated over AMQP 1.0 to swig client
> e.g. raises LinkError instead of NotFound if queue not found
> - auto-delete appears not to work with 1.0,
> - no TX support for 1.0.
> - Can't create an exclusive subscription over 1.0? This fails:
> "excl_sub;{create:always, link:{x-subscribe:{exclusive:True}}}");
>
>
> Diffs
> -----
>
> /trunk/qpid/cpp/bindings/qpid/python/qpid_messaging.i 1570699
> /trunk/qpid/cpp/include/qpid/messaging/Logger.h 1570699
> /trunk/qpid/cpp/include/qpid/qpid.i 1570699
> /trunk/qpid/cpp/include/qpid/swig_python_typemaps.i 1570699
> /trunk/qpid/cpp/src/amqp.cmake 1570699
> /trunk/qpid/cpp/src/qpid/messaging/ConnectionOptions.h 1570699
> /trunk/qpid/cpp/src/qpid/messaging/ConnectionOptions.cpp 1570699
> /trunk/qpid/cpp/src/qpid/messaging/amqp/AddressHelper.cpp 1570699
> /trunk/qpid/cpp/src/qpid/messaging/amqp/ConnectionContext.cpp 1570699
> /trunk/qpid/cpp/src/qpid/messaging/amqp/PnData.h PRE-CREATION
> /trunk/qpid/cpp/src/qpid/messaging/amqp/PnData.cpp PRE-CREATION
> /trunk/qpid/cpp/src/tests/CMakeLists.txt 1570699
> /trunk/qpid/cpp/src/tests/brokertest.py 1570699
> /trunk/qpid/cpp/src/tests/ha_test.py 1570699
> /trunk/qpid/cpp/src/tests/ha_tests.py 1570699
> /trunk/qpid/cpp/src/tests/interlink_tests.py 1570699
> /trunk/qpid/cpp/src/tests/qpidd_qmfv2_tests.py 1570699
> /trunk/qpid/cpp/src/tests/test_env.sh.in 1570699
> /trunk/qpid/tools/src/py/qpidtoollibs/broker.py 1570699
>
> Diff: https://reviews.apache.org/r/18387/diff/
>
>
> Testing
> -------
>
> ctest -VV with/without QPID_PY_NO_SWIG=1 on fedora 20 and rhel 6
>
>
> Thanks,
>
> Alan Conway
>
>
