ChangeLogTag:Thu Mar 23 10:39:23 2000 Carlos O'Ryan <coryan@uci.edu>

4 files changed, 119 insertions, 84 deletions

diff --git a/ChangeLog b/ChangeLog
index 60c19e53cda..3d6e7f6ab8f 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,19 +1,29 @@
+Thu Mar 23 10:39:23 2000  Carlos O'Ryan  <coryan@uci.edu>
+
+        * bin/g++dep:
+          Changed the script to generate relative paths for local
+          dependencies.  In other words if .obj/foo.o in the foo/bar
+          directory depends on $(ACE_ROOT)/foo/bar/foo.h we just generate
+          .obj/foo.o: foo.h
+          instead of the full path.  This solves some problems with
+          parallel builds and IDL generated files.
+
 Thu Mar 23 04:31:50 2000  Irfan Pyarali  <irfan@cs.wustl.edu>
 
-	* ace/Reactor.i (register_handler): There was a very subtle bug in
-	  all versions of <register_handler>.  The problem was that once
-	  the <event_handler> has been successfully registered with the
-	  Reactor, it is available for processing events.  In the problem
-	  case, input arrived from the client, the handler handled it, and
-	  eventually closed down.  All this happened in the other thread
-	  running the Reactor, and all before this line got executed:
+        * ace/Reactor.i (register_handler): There was a very subtle bug in
+          all versions of <register_handler>.  The problem was that once
+          the <event_handler> has been successfully registered with the
+          Reactor, it is available for processing events.  In the problem
+          case, input arrived from the client, the handler handled it, and
+          eventually closed down.  All this happened in the other thread
+          running the Reactor, and all before this line got executed:
 
-	      event_handler->reactor (this);
+              event_handler->reactor (this);
 
-	  By the time this thread came around to set the <reactor>, the
-	  <event_handler> had packed up and gone home. The fix is to set
-	  the <reactor> before registering the <event_handler> with the
-	  <reactor>.  Thanks to Steve Huston for report this bug.
+          By the time this thread came around to set the <reactor>, the
+          <event_handler> had packed up and gone home. The fix is to set
+          the <reactor> before registering the <event_handler> with the
+          <reactor>.  Thanks to Steve Huston for report this bug.
 
 Wed Mar 22 12:01:22 2000  Steve Huston  <shuston@riverace.com>
 
@@ -23,11 +33,11 @@ Wed Mar 22 12:01:22 2000  Steve Huston  <shuston@riverace.com>
 
 Wed Mar 22 10:19:47 2000  Douglas C. Schmidt  <schmidt@tango.cs.wustl.edu>
 
-	* ace/Thread.i: In ACE_Thread::setcancelstate() the code is
-	  checking to see if thr_setcancelstate and thr_setcanceltype
- 	  return 0 and if so returning an error condition to the user.
-  	  Instead, both of these routines should be checked to see if they
- 	  return -1 on failure.  Thanks to Umar Syyid for reporting this.
+        * ace/Thread.i: In ACE_Thread::setcancelstate() the code is
+          checking to see if thr_setcancelstate and thr_setcanceltype
+          return 0 and if so returning an error condition to the user.
+          Instead, both of these routines should be checked to see if they
+          return -1 on failure.  Thanks to Umar Syyid for reporting this.
 
 Wed Mar 22 09:29:09 2000  Chad Elliott  <elliott_c@ociweb.com>
 
@@ -36,20 +46,20 @@ Wed Mar 22 09:29:09 2000  Chad Elliott  <elliott_c@ociweb.com>
 
 Wed Mar 22 09:24:12 2000  Balachandran Natarajan  <bala@cs.wustl.edu>
 
-	* ace/Configuration.cpp: Fixed a link error with SunCC5. This
-	  error only shows up when ACE  is built with debug=0 and
-	  exceptions=1. The fix was to move all the template
-	  instantiations from the end of the file to the top of the
-	  file. Thanks to Dr.Levine for helping me fix this and thanks to
-	  Manuel Benche <mbenche@jazz.cs.utsa.edu> for reporting this
-	  problem.
+        * ace/Configuration.cpp: Fixed a link error with SunCC5. This
+          error only shows up when ACE  is built with debug=0 and
+          exceptions=1. The fix was to move all the template
+          instantiations from the end of the file to the top of the
+          file. Thanks to Dr.Levine for helping me fix this and thanks to
+          Manuel Benche <mbenche@jazz.cs.utsa.edu> for reporting this
+          problem.
 
 Wed Mar 22 09:16:48 2000  Jeff Parsons  <parsons@cs.wustl.edu>
 
-	* ace/CDR_Stream.i:
-	  Initialized a local variable inside the << operator
-	  for wstring. It was producing a warning on Linux g++
-	  (charanga).
+        * ace/CDR_Stream.i:
+          Initialized a local variable inside the << operator
+          for wstring. It was producing a warning on Linux g++
+          (charanga).
 
 Tue Mar 21 14:24:07 2000  Chad Elliott <elliott_c@ociweb.com>
 
diff --git a/ChangeLogs/ChangeLog-02a b/ChangeLogs/ChangeLog-02a
index 60c19e53cda..3d6e7f6ab8f 100644
--- a/ChangeLogs/ChangeLog-02a
+++ b/ChangeLogs/ChangeLog-02a
@@ -1,19 +1,29 @@
+Thu Mar 23 10:39:23 2000  Carlos O'Ryan  <coryan@uci.edu>
+
+        * bin/g++dep:
+          Changed the script to generate relative paths for local
+          dependencies.  In other words if .obj/foo.o in the foo/bar
+          directory depends on $(ACE_ROOT)/foo/bar/foo.h we just generate
+          .obj/foo.o: foo.h
+          instead of the full path.  This solves some problems with
+          parallel builds and IDL generated files.
+
 Thu Mar 23 04:31:50 2000  Irfan Pyarali  <irfan@cs.wustl.edu>
 
-	* ace/Reactor.i (register_handler): There was a very subtle bug in
-	  all versions of <register_handler>.  The problem was that once
-	  the <event_handler> has been successfully registered with the
-	  Reactor, it is available for processing events.  In the problem
-	  case, input arrived from the client, the handler handled it, and
-	  eventually closed down.  All this happened in the other thread
-	  running the Reactor, and all before this line got executed:
+        * ace/Reactor.i (register_handler): There was a very subtle bug in
+          all versions of <register_handler>.  The problem was that once
+          the <event_handler> has been successfully registered with the
+          Reactor, it is available for processing events.  In the problem
+          case, input arrived from the client, the handler handled it, and
+          eventually closed down.  All this happened in the other thread
+          running the Reactor, and all before this line got executed:
 
-	      event_handler->reactor (this);
+              event_handler->reactor (this);
 
-	  By the time this thread came around to set the <reactor>, the
-	  <event_handler> had packed up and gone home. The fix is to set
-	  the <reactor> before registering the <event_handler> with the
-	  <reactor>.  Thanks to Steve Huston for report this bug.
+          By the time this thread came around to set the <reactor>, the
+          <event_handler> had packed up and gone home. The fix is to set
+          the <reactor> before registering the <event_handler> with the
+          <reactor>.  Thanks to Steve Huston for report this bug.
 
 Wed Mar 22 12:01:22 2000  Steve Huston  <shuston@riverace.com>
 
@@ -23,11 +33,11 @@ Wed Mar 22 12:01:22 2000  Steve Huston  <shuston@riverace.com>
 
 Wed Mar 22 10:19:47 2000  Douglas C. Schmidt  <schmidt@tango.cs.wustl.edu>
 
-	* ace/Thread.i: In ACE_Thread::setcancelstate() the code is
-	  checking to see if thr_setcancelstate and thr_setcanceltype
- 	  return 0 and if so returning an error condition to the user.
-  	  Instead, both of these routines should be checked to see if they
- 	  return -1 on failure.  Thanks to Umar Syyid for reporting this.
+        * ace/Thread.i: In ACE_Thread::setcancelstate() the code is
+          checking to see if thr_setcancelstate and thr_setcanceltype
+          return 0 and if so returning an error condition to the user.
+          Instead, both of these routines should be checked to see if they
+          return -1 on failure.  Thanks to Umar Syyid for reporting this.
 
 Wed Mar 22 09:29:09 2000  Chad Elliott  <elliott_c@ociweb.com>
 
@@ -36,20 +46,20 @@ Wed Mar 22 09:29:09 2000  Chad Elliott  <elliott_c@ociweb.com>
 
 Wed Mar 22 09:24:12 2000  Balachandran Natarajan  <bala@cs.wustl.edu>
 
-	* ace/Configuration.cpp: Fixed a link error with SunCC5. This
-	  error only shows up when ACE  is built with debug=0 and
-	  exceptions=1. The fix was to move all the template
-	  instantiations from the end of the file to the top of the
-	  file. Thanks to Dr.Levine for helping me fix this and thanks to
-	  Manuel Benche <mbenche@jazz.cs.utsa.edu> for reporting this
-	  problem.
+        * ace/Configuration.cpp: Fixed a link error with SunCC5. This
+          error only shows up when ACE  is built with debug=0 and
+          exceptions=1. The fix was to move all the template
+          instantiations from the end of the file to the top of the
+          file. Thanks to Dr.Levine for helping me fix this and thanks to
+          Manuel Benche <mbenche@jazz.cs.utsa.edu> for reporting this
+          problem.
 
 Wed Mar 22 09:16:48 2000  Jeff Parsons  <parsons@cs.wustl.edu>
 
-	* ace/CDR_Stream.i:
-	  Initialized a local variable inside the << operator
-	  for wstring. It was producing a warning on Linux g++
-	  (charanga).
+        * ace/CDR_Stream.i:
+          Initialized a local variable inside the << operator
+          for wstring. It was producing a warning on Linux g++
+          (charanga).
 
 Tue Mar 21 14:24:07 2000  Chad Elliott <elliott_c@ociweb.com>
 
diff --git a/ChangeLogs/ChangeLog-03a b/ChangeLogs/ChangeLog-03a
index 60c19e53cda..3d6e7f6ab8f 100644
--- a/ChangeLogs/ChangeLog-03a
+++ b/ChangeLogs/ChangeLog-03a
@@ -1,19 +1,29 @@
+Thu Mar 23 10:39:23 2000  Carlos O'Ryan  <coryan@uci.edu>
+
+        * bin/g++dep:
+          Changed the script to generate relative paths for local
+          dependencies.  In other words if .obj/foo.o in the foo/bar
+          directory depends on $(ACE_ROOT)/foo/bar/foo.h we just generate
+          .obj/foo.o: foo.h
+          instead of the full path.  This solves some problems with
+          parallel builds and IDL generated files.
+
 Thu Mar 23 04:31:50 2000  Irfan Pyarali  <irfan@cs.wustl.edu>
 
-	* ace/Reactor.i (register_handler): There was a very subtle bug in
-	  all versions of <register_handler>.  The problem was that once
-	  the <event_handler> has been successfully registered with the
-	  Reactor, it is available for processing events.  In the problem
-	  case, input arrived from the client, the handler handled it, and
-	  eventually closed down.  All this happened in the other thread
-	  running the Reactor, and all before this line got executed:
+        * ace/Reactor.i (register_handler): There was a very subtle bug in
+          all versions of <register_handler>.  The problem was that once
+          the <event_handler> has been successfully registered with the
+          Reactor, it is available for processing events.  In the problem
+          case, input arrived from the client, the handler handled it, and
+          eventually closed down.  All this happened in the other thread
+          running the Reactor, and all before this line got executed:
 
-	      event_handler->reactor (this);
+              event_handler->reactor (this);
 
-	  By the time this thread came around to set the <reactor>, the
-	  <event_handler> had packed up and gone home. The fix is to set
-	  the <reactor> before registering the <event_handler> with the
-	  <reactor>.  Thanks to Steve Huston for report this bug.
+          By the time this thread came around to set the <reactor>, the
+          <event_handler> had packed up and gone home. The fix is to set
+          the <reactor> before registering the <event_handler> with the
+          <reactor>.  Thanks to Steve Huston for report this bug.
 
 Wed Mar 22 12:01:22 2000  Steve Huston  <shuston@riverace.com>
 
@@ -23,11 +33,11 @@ Wed Mar 22 12:01:22 2000  Steve Huston  <shuston@riverace.com>
 
 Wed Mar 22 10:19:47 2000  Douglas C. Schmidt  <schmidt@tango.cs.wustl.edu>
 
-	* ace/Thread.i: In ACE_Thread::setcancelstate() the code is
-	  checking to see if thr_setcancelstate and thr_setcanceltype
- 	  return 0 and if so returning an error condition to the user.
-  	  Instead, both of these routines should be checked to see if they
- 	  return -1 on failure.  Thanks to Umar Syyid for reporting this.
+        * ace/Thread.i: In ACE_Thread::setcancelstate() the code is
+          checking to see if thr_setcancelstate and thr_setcanceltype
+          return 0 and if so returning an error condition to the user.
+          Instead, both of these routines should be checked to see if they
+          return -1 on failure.  Thanks to Umar Syyid for reporting this.
 
 Wed Mar 22 09:29:09 2000  Chad Elliott  <elliott_c@ociweb.com>
 
@@ -36,20 +46,20 @@ Wed Mar 22 09:29:09 2000  Chad Elliott  <elliott_c@ociweb.com>
 
 Wed Mar 22 09:24:12 2000  Balachandran Natarajan  <bala@cs.wustl.edu>
 
-	* ace/Configuration.cpp: Fixed a link error with SunCC5. This
-	  error only shows up when ACE  is built with debug=0 and
-	  exceptions=1. The fix was to move all the template
-	  instantiations from the end of the file to the top of the
-	  file. Thanks to Dr.Levine for helping me fix this and thanks to
-	  Manuel Benche <mbenche@jazz.cs.utsa.edu> for reporting this
-	  problem.
+        * ace/Configuration.cpp: Fixed a link error with SunCC5. This
+          error only shows up when ACE  is built with debug=0 and
+          exceptions=1. The fix was to move all the template
+          instantiations from the end of the file to the top of the
+          file. Thanks to Dr.Levine for helping me fix this and thanks to
+          Manuel Benche <mbenche@jazz.cs.utsa.edu> for reporting this
+          problem.
 
 Wed Mar 22 09:16:48 2000  Jeff Parsons  <parsons@cs.wustl.edu>
 
-	* ace/CDR_Stream.i:
-	  Initialized a local variable inside the << operator
-	  for wstring. It was producing a warning on Linux g++
-	  (charanga).
+        * ace/CDR_Stream.i:
+          Initialized a local variable inside the << operator
+          for wstring. It was producing a warning on Linux g++
+          (charanga).
 
 Tue Mar 21 14:24:07 2000  Chad Elliott <elliott_c@ociweb.com>
 
diff --git a/bin/g++dep b/bin/g++dep
index 823d29b0776..3acbfb15929 100755
--- a/bin/g++dep
+++ b/bin/g++dep
@@ -51,6 +51,11 @@ case $1 in
 esac
 done
 
+# Dependencies on local files are better expressed like that, instead
+# of using $(TAO_ROOT) or $(ACE_ROOT).  This is specially important
+# for IDL generated files.
+REL="-e s;$PWD/;;g "$REL
+
 if [ ! -w $MAKE ]; then
   echo "g++dep: no writeable file \"$MAKE\""
   exit 1