| Summary: | Implement exception handling for Asynchronous Messaging | ||
|---|---|---|---|
| Product: | TAO | Reporter: | Carlos O'Ryan <coryan> |
| Component: | ORB | Assignee: | Michael Kircher <Michael.Kircher> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | ||
| Priority: | P2 | ||
| Version: | 1.0.2 | ||
| Hardware: | All | ||
| OS: | All | ||
| Bug Depends on: | 175 | ||
| Bug Blocks: | |||
|
Description
Carlos O'Ryan
1999-08-26 10:30:06 CDT
I will accept this one but need to discuss this with Carlos. Guess this is an important feature thats missing and hence almost renders AMI unusable!!! The last comment is not completely accurate (or helpful), exception handling does work, the only case that fails is when the server crashes during an AMI invocation. Maybe Michael can fix this up? I will look into it and see what I can do. Exception handling in general is available now, but I guess we do not inform the Reply Handler if we are unable to send the request. Carlos suggested: What we should do is send a message to the transport mux strategy, saying something like "your connection is going down", the transport mux strategy can send a message to each one of the Reply_Dispatchers, saying "the connection is down". The Synch reply dispatchers can then simply set a flag, wake up the waiting thread and then the waiting thread raises COMM_FAILURE. The ASynch reply dispatchers can invoke the ReplyHandler with the correct exception too. During my fixes to bug 175 I discovered that this two are related. I will try to solve as much of this bug (266) as I go along 175 The bug as originally described has been fixed. AMI callbacks still have problems with CloseConnection messages, as described in bug 124 It should have been marked fixed in the last change. |