ChangeLogTag:Wed May 2 18:36:29 2001 Douglas C. Schmidt <schmidt@tango.doc.wustl.edu>

8 files changed, 141 insertions, 53 deletions

diff --git a/ChangeLog b/ChangeLog
index e04529ffeff..c9ef0fbe012 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,9 @@
+Wed May  2 18:36:29 2001  Douglas C. Schmidt  <schmidt@tango.doc.wustl.edu>
+
+	* tests/README: Clarified that the shared memory tests now
+	  work fine on Win9x!  Thanks to Cristian Ferritti for reminding
+ 	  me to do this.
+
 Wed May  2 15:10:28 2001  Balachandran Natarajan  <bala@cs.wustl.edu>
 
 	* ace/CDR_Stream.h (ACE_InputCDR): 
@@ -18,9 +24,23 @@ Wed May 02 05:40:47 2001  Carlos O'Ryan  <coryan@uci.edu>
         * bin/auto_run_tests.lst:
           Use Queued_Message_Test/run_test.pl
 
-Tue May  1 21:38:00 2001  Balachandran Natarajan  <bala@cs.wustl.edu>
-
-        * THANKS: Added Carlton Teel to the hall of fame.
+Wed May  2 06:31:46 2001  Douglas C. Schmidt  <schmidt@tango.doc.wustl.edu>
+
+	* ace/Makefile: Make it possible to set where the QTDIR is located.
+	  Thanks to Thomas Groth <groth.th@nord-com.net> for this
+ 	  enhancement.
+
+	* ace/Process.cpp (spawn): When an application running with user
+	  ID=0 (i.e. root) spawns a new process using
+ 	  ACE_Process_Manager::spawn(), which in turn calls
+ 	  ACE_Process::spawn() and tries to set a user ID and a group ID
+ 	  for the new process different from root/other, the call to
+ 	  setreuid() succeedes but the call to setregid() made thereafter
+ 	  fails, because the process has no longer UID=0 and thus it
+ 	  cannot change to a different group.  To fix this, it was
+ 	  necessary to swap the order of the calls to setreuid() and
+ 	  setregid().  Thanks to Ramiro Penataro <penataro@ll.iac.es> for
+ 	  reporting this.
 
 Tue May  1 11:05:48 2001  Ossama Othman  <ossama@uci.edu>
 
diff --git a/ChangeLogs/ChangeLog-02a b/ChangeLogs/ChangeLog-02a
index e04529ffeff..c9ef0fbe012 100644
--- a/ChangeLogs/ChangeLog-02a
+++ b/ChangeLogs/ChangeLog-02a
@@ -1,3 +1,9 @@
+Wed May  2 18:36:29 2001  Douglas C. Schmidt  <schmidt@tango.doc.wustl.edu>
+
+	* tests/README: Clarified that the shared memory tests now
+	  work fine on Win9x!  Thanks to Cristian Ferritti for reminding
+ 	  me to do this.
+
 Wed May  2 15:10:28 2001  Balachandran Natarajan  <bala@cs.wustl.edu>
 
 	* ace/CDR_Stream.h (ACE_InputCDR): 
@@ -18,9 +24,23 @@ Wed May 02 05:40:47 2001  Carlos O'Ryan  <coryan@uci.edu>
         * bin/auto_run_tests.lst:
           Use Queued_Message_Test/run_test.pl
 
-Tue May  1 21:38:00 2001  Balachandran Natarajan  <bala@cs.wustl.edu>
-
-        * THANKS: Added Carlton Teel to the hall of fame.
+Wed May  2 06:31:46 2001  Douglas C. Schmidt  <schmidt@tango.doc.wustl.edu>
+
+	* ace/Makefile: Make it possible to set where the QTDIR is located.
+	  Thanks to Thomas Groth <groth.th@nord-com.net> for this
+ 	  enhancement.
+
+	* ace/Process.cpp (spawn): When an application running with user
+	  ID=0 (i.e. root) spawns a new process using
+ 	  ACE_Process_Manager::spawn(), which in turn calls
+ 	  ACE_Process::spawn() and tries to set a user ID and a group ID
+ 	  for the new process different from root/other, the call to
+ 	  setreuid() succeedes but the call to setregid() made thereafter
+ 	  fails, because the process has no longer UID=0 and thus it
+ 	  cannot change to a different group.  To fix this, it was
+ 	  necessary to swap the order of the calls to setreuid() and
+ 	  setregid().  Thanks to Ramiro Penataro <penataro@ll.iac.es> for
+ 	  reporting this.
 
 Tue May  1 11:05:48 2001  Ossama Othman  <ossama@uci.edu>
 
diff --git a/ChangeLogs/ChangeLog-03a b/ChangeLogs/ChangeLog-03a
index e04529ffeff..c9ef0fbe012 100644
--- a/ChangeLogs/ChangeLog-03a
+++ b/ChangeLogs/ChangeLog-03a
@@ -1,3 +1,9 @@
+Wed May  2 18:36:29 2001  Douglas C. Schmidt  <schmidt@tango.doc.wustl.edu>
+
+	* tests/README: Clarified that the shared memory tests now
+	  work fine on Win9x!  Thanks to Cristian Ferritti for reminding
+ 	  me to do this.
+
 Wed May  2 15:10:28 2001  Balachandran Natarajan  <bala@cs.wustl.edu>
 
 	* ace/CDR_Stream.h (ACE_InputCDR): 
@@ -18,9 +24,23 @@ Wed May 02 05:40:47 2001  Carlos O'Ryan  <coryan@uci.edu>
         * bin/auto_run_tests.lst:
           Use Queued_Message_Test/run_test.pl
 
-Tue May  1 21:38:00 2001  Balachandran Natarajan  <bala@cs.wustl.edu>
-
-        * THANKS: Added Carlton Teel to the hall of fame.
+Wed May  2 06:31:46 2001  Douglas C. Schmidt  <schmidt@tango.doc.wustl.edu>
+
+	* ace/Makefile: Make it possible to set where the QTDIR is located.
+	  Thanks to Thomas Groth <groth.th@nord-com.net> for this
+ 	  enhancement.
+
+	* ace/Process.cpp (spawn): When an application running with user
+	  ID=0 (i.e. root) spawns a new process using
+ 	  ACE_Process_Manager::spawn(), which in turn calls
+ 	  ACE_Process::spawn() and tries to set a user ID and a group ID
+ 	  for the new process different from root/other, the call to
+ 	  setreuid() succeedes but the call to setregid() made thereafter
+ 	  fails, because the process has no longer UID=0 and thus it
+ 	  cannot change to a different group.  To fix this, it was
+ 	  necessary to swap the order of the calls to setreuid() and
+ 	  setregid().  Thanks to Ramiro Penataro <penataro@ll.iac.es> for
+ 	  reporting this.
 
 Tue May  1 11:05:48 2001  Ossama Othman  <ossama@uci.edu>
 
diff --git a/README b/README
index 9d862053b62..ef6034bc58e 100644
--- a/README
+++ b/README
@@ -221,5 +221,5 @@ To ensure that you see responses, please do one of the following:
 
 ACKNOWLEDGEMENTS
 
-Please see the file `THANKS' for a list of people who have contributed
-to ACE and TAO over the years.
+Please see the file `$ACE_ROOT/THANKS' for a list of the thousands of
+people who've contributed to ACE and TAO over the years.
diff --git a/TAO/ChangeLogs/ChangeLog-02a b/TAO/ChangeLogs/ChangeLog-02a
index 7496ee89c41..43a445f9126 100644
--- a/TAO/ChangeLogs/ChangeLog-02a
+++ b/TAO/ChangeLogs/ChangeLog-02a
@@ -63,16 +63,15 @@ Wed May  2 11:26:10 2001  Jeff Parsons <parsons@cs.wustl.edu>
 
         * examples/POA/TIE/Makefile:
 
-          Disabled this build for SunCC 4.2. It worked before, even
-          though 4.2 does not have ACE_HAS_USING_KEYWORD defined
-          (in other words, it has no namespace support), because the
-          IDL compiler left out the #ifdef guards using the above
-          macro if the interface was declared at global scope, and the
-          only interface in the TIE example where the tie class was
-          used happened to fall into that category. This generation of
-          #ifdef guards was inconsistent, however, and now *all* tie
-          classes are left out by the preprocessor if 
-          ACE_HAS_USING_KEYWORD is not defined.
+          Disabled this build for SunCC 4.2. It worked before, even though
+ 	  4.2 does not have ACE_HAS_USING_KEYWORD defined (in other words,
+ 	  it has no namespace support), because the IDL compiler left out
+ 	  the #ifdef guards using the above macro if the interface was
+ 	  declared at global scope, and the only interface in the TIE
+ 	  example where the tie class was used happened to fall into that
+ 	  category. This generation of #ifdef guards was inconsistent,
+ 	  however, and now *all* tie classes are left out by the
+ 	  preprocessor if ACE_HAS_USING_KEYWORD is not defined.
 
 Wed May 02 05:37:23 2001  Carlos O'Ryan  <coryan@uci.edu>
 
@@ -84,8 +83,8 @@ Tue May 1 22:42:44 2001  Ossama Othman  <ossama@uci.edu>
 
         * tao/GIOP_Message_Generator_Parser_10.cpp (write_reply_header):
 
-          Silence the false Fuzz error about a missing ACE_CHECK by adding
-          the awful "// FUZZ: ignore check_for_ace_check" comment hack.
+	  Silence the false Fuzz error about a missing ACE_CHECK by adding
+ 	  the awful "// FUZZ: ignore check_for_ace_check" comment hack.
 
 Tue May  1 22:14:01 2001  Balachandran Natarajan  <bala@cs.wustl.edu>
 
@@ -289,9 +288,36 @@ Mon Apr 30 16:53:27 2001  Jeff Parsons <parsons@cs.wustl.edu>
 
 Mon Apr 30 11:23:45 2001  Ossama Othman  <ossama@uci.edu>
 
+	* tao/Cache_Entries.h (relinquish_transport):
+	* tao/Cache_Entries.inl (relinquish_transport):
+
+	  New method that causes the given TAO_Cache_IntId to relinquish
+	  ownership of the TAO_Transport associated with it.  This method
+	  exists to get around some inflexibility in the
+	  Transport_Cache_Manager interface.  It should be removed once
+	  the interface is improved.
+
+	* tao/Transport_Cache_Manager.cpp (find_transport):
+
+	  Do not duplicate the TAO_Transport object after it is found.
+	  It is already duplicated during the find process.  Instead call
+	  the new TAO_Cache_IntId::relinquish_transport() method.  This
+	  change removes two locks from the critical path.  [Bug 878]
+
+<<<<<<< ChangeLog
+	* tao/IIOP_Endpoint.cpp (hash):
+=======
         * tao/Cache_Entries.h (relinquish_transport):
         * tao/Cache_Entries.inl (relinquish_transport):
-
+>>>>>>> 1.3407
+
+<<<<<<< ChangeLog
+	  Improved the speed of this hash operation by basing the hash on
+	  the IP address of the peer instead of computing a hash that
+	  involves some string operations (ACE::hash_pjw()).  Since this
+	  hash method is called on each invocation, i.e. in the critical
+	  path, a performance improvement should be noticeable.
+=======
           New method that causes the given TAO_Cache_IntId to relinquish
           ownership of the TAO_Transport associated with it.  This method
           exists to get around some inflexibility in the
@@ -312,6 +338,7 @@ Mon Apr 30 11:23:45 2001  Ossama Othman  <ossama@uci.edu>
           involves some string operations (ACE::hash_pjw()).  Since this
           hash method is called on each invocation, i.e. in the critical
           path, a performance improvement should be noticeable.
+>>>>>>> 1.3407
 
         * tao/PortableServer/Servant_Base.cpp:
 
@@ -16409,10 +16436,10 @@ Fri Dec 22 14:13:46 2000  Priyanka Gontla  <pgontla@ece.uci.edu>
           RT_Current, Priority_Mapping_Manager and RT_Policy_i.
 
           Removed the set_default_policies method. It now is in
-          Protocols_Hooks and derived classes. The default implementation
-          does nothing and the RT_Protocols_Hooks implementation does the
-          actual setting of the policies. With this change, ORB_Core doesnt
-          depend on RT_Policy_i.
+ 	  Protocols_Hooks and derived classes. The default implementation
+ 	  does nothing and the RT_Protocols_Hooks implementation does the
+ 	  actual setting of the policies. With this change, ORB_Core
+ 	  doesnt depend on RT_Policy_i.
 
           Similar is the case with get_thread_priority and
           set_thread_priority methods.
@@ -19691,10 +19718,10 @@ Wed Nov 22 16:24:52 2000  Ossama Othman  <ossama@uci.edu>
 
         * orbsvcs/orbsvcs/SSLIOP/SSLIOP_Connector.cpp (connect):
 
-          Corrected typo where base class pointer was checked for
-          validity instead of the derived class pointer.  This fixes a
-          potential problem where a failed dynamic_cast would not have
-          beend etected.
+          Corrected typo where base class pointer was checked for validity
+ 	  instead of the derived class pointer.  This fixes a potential
+ 	  problem where a failed dynamic_cast would not have beend
+ 	  etected.
 
         * orbsvcs/orbsvcs/SSLIOP/SSLIOP_Endpoint.cpp
           (TAO_SSLIOP_Endpoint):
diff --git a/ace/Makefile b/ace/Makefile
index ad82cc1f3ef..54371579a00 100644
--- a/ace/Makefile
+++ b/ace/Makefile
@@ -562,12 +562,18 @@ Svc_Conf_l.cpp: Svc_Conf.l
         fi
 
 # Some compilation for the QtReactor stuff. We need to run moc
-# ie. (Meta Object compiler) for our QtReactor.h. So, let me do it
+# ie. (Meta Object compiler) for our QtReactor.h. So, let's do it
 # here.
 
+ifneq ($(QTDIR),)
+  MOC = ${QTDIR}/bin/moc
+else
+  MOC = moc
+endif
+
 ifneq ($(qt_reactor),)
 QtReactor_moc.cpp: QtReactor.h
-	moc QtReactor.h >QtReactor_moc.cpp
+	$(MOC) QtReactor.h >QtReactor_moc.cpp
 endif #qt_reactor
 
 ifeq ($(SUPPRESS_DASH_G),1)
diff --git a/ace/Process.cpp b/ace/Process.cpp
index f4df495b037..050391ef389 100644
--- a/ace/Process.cpp
+++ b/ace/Process.cpp
@@ -125,6 +125,16 @@ ACE_Process::spawn (ACE_Process_Options &options)
                     ACE_LIB_TEXT ("%p.\n"),
                     ACE_LIB_TEXT ("ACE_Process::spawn: setpgid failed.")));
 
+#if !defined (ACE_LACKS_SETREGID)
+      if (options.getrgid () != (uid_t) -1
+          || options.getegid () != (uid_t) -1)
+        if (ACE_OS::setregid (options.getrgid (),
+                              options.getegid ()) == -1)
+          ACE_ERROR ((LM_ERROR,
+                      ACE_LIB_TEXT ("%p.\n"),
+                      ACE_LIB_TEXT ("ACE_Process::spawn: setregid failed.")));
+#endif /* ACE_LACKS_SETREGID */
+
 #if !defined (ACE_LACKS_SETREUID)
       // Set user and group id's.
       if (options.getruid () != (uid_t) -1
@@ -136,16 +146,6 @@ ACE_Process::spawn (ACE_Process_Options &options)
                       ACE_LIB_TEXT ("ACE_Process::spawn: setreuid failed.")));
 #endif /* ACE_LACKS_SETREUID */
 
-#if !defined (ACE_LACKS_SETREGID)
-      if (options.getrgid () != (uid_t) -1
-          || options.getegid () != (uid_t) -1)
-        if (ACE_OS::setregid (options.getrgid (),
-                              options.getegid ()) == -1)
-          ACE_ERROR ((LM_ERROR,
-                      ACE_LIB_TEXT ("%p.\n"),
-                      ACE_LIB_TEXT ("ACE_Process::spawn: setregid failed.")));
-#endif /* ACE_LACKS_SETREGID */
-
       this->child (ACE_OS::getppid ());
     }
   else if (this->child_id_ != -1)
diff --git a/tests/README b/tests/README
index faef89c56c8..6cb0502b965 100644
--- a/tests/README
+++ b/tests/README
@@ -34,26 +34,21 @@ The tests have been run on a wide range of platforms (e.g., UNIX,
 pSOS, VxWorks, LynxOS, Windows NT/2000/95/98/etc.) and they all work
 with the following exceptions:
 
-1. Naming_Test, MEM_Stream_Test, Malloc_Test, and
-   MM_Shared_Memory_Test on Windows 98/95/ME: these do not work on
-   Windows 9x since Windows 9x has a different model for shared memory
-   and does not support explicitly selecting shared memory addresses.
-
-2. UPIPE_Test and SPIPE_Test on Windows 95: these do not work on
+1. UPIPE_Test and SPIPE_Test on Windows 95: these tests don't run on
    Windows 95 since Windows 95 does not support the server side
    functionality of accepting connections from clients using named pipes.
 
-3. Proactor_Test (to be added) on Windows 95: this does not work on
+2. Proactor_Test (to be added) on Windows 95: this test does not run on
    Windows 95 since Windows 95 does not support I/O completion ports.
 
-4. Of the tests (believed) relevant to the single threaded port to pSOSim,
+3. Of the tests (believed) relevant to the single threaded port to pSOSim,
    Sigset_Ops_Test, Sock_Connector_Test, Priority_Reactor_Test,
    Process_Strategy_Test, and Service_Config_Test are still failing.
    These are being debugged and will either be corrected or, if there
    are unsupported features in these tests, removed from the test suite for
    pSOSim.
 
-5. The set of tests for pSOSim will be expanded to include those that require
+4. The set of tests for pSOSim will be expanded to include those that require
    multiple threads, with the release of the multi-threaded port to pSOSim.
 
 Notes: