| Summary: | Client side is not handling CloseConnection GIOP message properly | ||
|---|---|---|---|
| Product: | TAO | Reporter: | Ossama Othman <ossama.othman> |
| Component: | ORB | Assignee: | Chris Cleeland <cleeland_c> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | ||
| Priority: | P3 | ||
| Version: | 1.0 | ||
| Hardware: | All | ||
| OS: | All | ||
| Bug Depends on: | |||
| Bug Blocks: | 1444 | ||
|
Description
Ossama Othman
1999-07-26 17:17:45 CDT
I'll be point man for this bug. Simply accepting the bugs. I don't have time to work in this back at any point in the foreseeable future. I'm returning the bug to the main pool, if anybody wants to take it they are welcomed to ask questions on how to approach the problem/enhancement. Bugs moved to the tao-support placeholder. The client side ignores the CloseConnection messages, and all waiting replies raise the COMM_FAILURE exception. The right behavior is to raise the COMM_FAILURE exception only if the connection is closed without an intervening CloseConnection message. Chris, can you have a look if this is still valid, you fixed some issues related to CloseConnection earlier this year (Finally getting to this!) Best that I can tell, proper handling of receipt of GIOP CloseConnection is performed. This was committed as part of this changelog entry: Wed Jan 3 05:58:08 UTC 2007 Chris Cleeland <cleeland_c@ociweb.com> * tao/GIOP_Message_State.cpp (parse_message_header_i): Fixed a problem where TAO mistakenly considered ALL messages with zero-length payload to be errors and was thus not properly parsing and handling the GIOP CloseConnection message. * tao/GIOP_Utils.h: Corrected comment incorrectly indicating that CloseConnection could not be send by both sides of a connection. * tests/Bug_2702_Regression/CloseConnection.mpc: * tests/Bug_2702_Regression/README: * tests/Bug_2702_Regression/Test.idl: * tests/Bug_2702_Regression/client.cpp: * tests/Bug_2702_Regression/closeconnection.dat: * tests/Bug_2702_Regression/fakeserver.pm: * tests/Bug_2702_Regression/fakeserver2.pl: * tests/Bug_2702_Regression/run_test.pl: Added a test to simulate the receipt of a CloseConnection in a client, similar to what the original reporter experienced. See the accompanying README for details. I *think*, then, that this bug could be marked closed. The bug itself is pretty vague, so it's hard to decide exactly what the criteria are for closing it. I leave that to the original submitter. closing as fixed |