diff options
Diffstat (limited to 'TAO/ChangeLog')
| -rw-r--r-- | TAO/ChangeLog | 132 |
1 files changed, 120 insertions, 12 deletions
diff --git a/TAO/ChangeLog b/TAO/ChangeLog index 7f3ef65cb48..9d85cd87c03 100644 --- a/TAO/ChangeLog +++ b/TAO/ChangeLog @@ -1,3 +1,114 @@ +Thu Oct 27 02:53:53 UTC 2006 Ossama Othman <ossama_othman at symantec dot com> + + * tao/GIOP_Message_Base.cpp (parse_request_id): + + Corrected GIOP version check so that the GIOP 1.{0,1} case + doesn't apply to all major versions greater than or equal to + one, i.e. {1,2,3,...,N}.{0,1}. The case in question is only + meant for GIOP 1.0 or 1.1. + +Thu Oct 27 01:59:19 UTC 2006 Ossama Othman <ossama_othman at symantec dot com> + + * tao/GIOP_Message_Base.cpp (parse_request_id): + + Simplified error return logic. Addresses "code not reached" + warning. + + * tao/ORB.cpp (ORB_init): + + Do not pass down the default CORBA::Environment from TSS in the + native C++ exception case. Simply pass a locally instantiated + CORBA::Environment. Removes an unnecessary TSS access. + + * tao/EndpointPolicy/Endpoint_Acceptor_Filter.h: + + Moved TAO_POA_Manager forward declaration into the versioned + namespace. + + * tao/PortableServer/Servant_var.h: + + Removed suggestion of adding an exception specification. + They've fallen out of favor by most C++ gurus. + + * tao/Strategies/advanced_resource.cpp: + * tao/Strategies/advanced_resource.h: + + Added support ACE_Dev_Poll_Reactor. + + * tao/Valuetype/ValueBase.cpp (_tao_write_value_header): + + Fixed Coverity DEADCODE error. + + From Russell Mora <russell_mora at symantec dot com> + * examples/Content_Server/AMI_Iterator/Content_Iterator_i.cpp: + * examples/Content_Server/SMI_Iterator/Content_Iterator_i.cpp: + * tao/Strategies/SHMIOP_Acceptor.cpp: + * tao/Strategies/SHMIOP_Acceptor.h: + * tao/Strategies/SHMIOP_Factory.h: + + Added support for 64-bit file offsets. + + From Duane Binder <duane_binder at symantec dot com> + * TAO_IDL/be/be_helper.cpp: + + Buffer I/O by removing fflush() calls. Greatly improves tao_idl + performance, particularly over networked filesystems. + + * TAO_IDL/be/be_interface.cpp (gen_gperf_lookup_methods): + + Flush the output stream. Gperf also uses it as output. Ensure + current contents are written before gperf writes. + +Thu Oct 26 14:20:12 UTC 2006 Johnny Willemsen <jwillemsen@remedy.nl> + + * orbsvcs/tests/Event/UDP/receiver.cpp: + Register the value factory + + * orbsvcs/tests/Event/UDP/run_test.pl: + Also run the test with a valuetype in the event. + +Thu Oct 26 13:30:12 UTC 2006 Johnny Willemsen <jwillemsen@remedy.nl> + + * orbsvcs/orbsvcs/Event/ECG_CDR_Message_Receiver.cpp: + Reverted the change below. Thew new UDP RTEC test showed + that the consumer was receiving duplicate events without being + able to detect that. So a single send resulted in two events + being received. The original problem needs to be resolved in + a different way because this breaks any assumptions about events + + Mon Aug 21 15:37:23 UTC 2006 Douglas C. Schmidt <schmidt@dre.vanderbilt.edu> + * orbsvcs/orbsvcs/Event/ECG_UDP_Out_Endpoint.i: + * orbsvcs/orbsvcs/Event/ECG_CDR_Message_Receiver.cpp: Made the + simplest fix to the problem where if a UDP RTEC supplier + restarts while the consumer is still running, it gets duplicate + fragment errors. The cost of this fix is that duplicate + completed requests are no longer detected. Thanks to Keith + Nicewarner for these fixes. + +Thu Oct 26 12:54:12 UTC 2006 Johnny Willemsen <jwillemsen@remedy.nl> + + * orbsvcs/orbsvcs/Event/ECG_CDR_Message_Sender.cpp: + * orbsvcs/orbsvcs/Event/ECG_Mcast_EH.cpp: + Const improvements + +Thu Oct 26 10:53:12 UTC 2006 Johnny Willemsen <jwillemsen@remedy.nl> + + * orbsvcs/tests/Event/UDP/*: + New regression test for the RTEC using UDP federations. By default + we just send the string ACE/TAO/CIAO in the Any and this works, but + when -v is passed to the run_test.pl we send a valuetype which + contains the string. With the current svn head version of today + this fails because on the receiver side we can't extract the valuetype + out of the Any + +Wed Oct 25 15:33:27 UTC 2006 Chad Elliott <elliott_c@ociweb.com> + + * tao/PortableServer/Root_POA.cpp: + + My change from Mon Oct 23 12:24:57 UTC 2006 uncovered yet another + reference counting problem. The TAO_Root_POA was not manipulating + the reference count of the TAO_POA_Manager that it held. + Mon Oct 23 12:24:57 UTC 2006 Chad Elliott <elliott_c@ociweb.com> * tao/PortableServer/Object_Adapter.h: @@ -102,14 +213,14 @@ Mon Oct 16 21:16:14 UTC 2006 Phil Mesnier <mesnier_p@ociweb.com> Mon Oct 16 10:59:12 UTC 2006 Johnny Willemsen <jwillemsen@remedy.nl> - * tests/Nested_Upcall_Crash/run_test.pl: - Check whether we could spawn the server, if not, give an error - and exit directly. + * tests/Nested_Upcall_Crash/run_test.pl: + Check whether we could spawn the server, if not, give an error + and exit directly. Mon Oct 16 08:43:12 UTC 2006 Johnny Willemsen <jwillemsen@remedy.nl> - * TAO_IDL/be/be_codegen.cpp: - Const changes + * TAO_IDL/be/be_codegen.cpp: + Const changes Sun Oct 15 23:23:53 UTC 2006 Phil Mesnier <mesnier_p@ociweb.com> @@ -132,15 +243,12 @@ Sun Oct 15 23:23:53 UTC 2006 Phil Mesnier <mesnier_p@ociweb.com> forces the race condition. Adding a small sleep along with the above fix demonstrates that the problem no longer occurs. - Thanks to Chris Reed, cr at progress dot com, for reporting this - bug. - Fri Oct 13 10:28:12 2006 Johnny Willemsen <jwillemsen@remedy.nl> - * tests/COIOP/*: - Extended this test so that we check whether a call should - succeed or not. This test requires that COIOP is the only - pluggable protocol available. + * tests/COIOP/*: + Extended this test so that we check whether a call should + succeed or not. This test requires that COIOP is the only + pluggable protocol available. Thu Oct 12 14:17:11 UTC 2006 Chad Elliott <elliott_c@ociweb.com> |