Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Fixed
-
0.27
-
None
-
Windows .NET binding
Description
Running various client invalid URL tests gets different results on amqp 0-10 vs 1.0.
Both versions get into Url::parse and throw Url::Invalid();
Under 0-10 (which works as expected) has
void ConnectionImpl::open() { qpid::sys::AbsTime start = qpid::sys::now(); qpid::sys::ScopedLock<qpid::sys::Semaphore> l(semaphore); try { if (!connection.isOpen()) connect(start); } catch (const types::Exception&) { throw; } catch (const qpid::Exception& e) { throw messaging::ConnectionError(e.what()); } }
where the Url throw is caught as a qpid::Exception and rethrown as a messaging::ConnectionError.
Under amqp1.0 the Url::Invalid bleeds up to the .NET client interface and is not handled correctly.
Maybe there are two things to fix:
1. Make the amqp1.0 client send messaging:: exceptions only.
2. Beef up the .NET binding to catch various other exceptions and relay them into the .NET client when possible.
As is the .NET clients see an exception of InteropServices::SEHUnhandledException which offers no hint of what went wrong.