diff options
Diffstat (limited to 'ACE/ChangeLog')
| -rw-r--r-- | ACE/ChangeLog | 3014 |
1 files changed, 3014 insertions, 0 deletions
diff --git a/ACE/ChangeLog b/ACE/ChangeLog index eac65d27ca5..7354bcd6be2 100644 --- a/ACE/ChangeLog +++ b/ACE/ChangeLog @@ -1,3 +1,3017 @@ + +Wed Mar 21 12:12:26 CDT 2007 Chad Elliott <elliott_c@ociweb.com> + + * ACE version 5.5.7 released. + +Tue Mar 20 12:42:13 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/diff-builds.pl: + + Sort the builds so that the output is in order and comparable + between multiple runs of the same date. + +Mon Mar 19 19:16:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/tao_other_tests.lst: + Disabled several tests for CORBA/e micro + +Mon Mar 19 16:37:21 UTC 2007 Douglas C. Schmidt <schmidt@dre.vanderbilt.edu> + + * docs/Download.html: Updated the comments about commercial + support so they point to www.dre.vanderbilt.edu/support.html. + Thanks to Hans van't Hag <hans dot vanthag at prismtech dot com> + for motivating this. + +Mon Mar 12 16:19:49 UTC 2007 Douglas C. Schmidt <schmidt@dre.vanderbilt.edu> + + * tests/Future_Test.cpp: Updated this example to point to the + right Active Object location. Thanks to Joe French <JFrench at + Soneticom dot com> for motivating this. + +Mon Mar 19 07:53:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/MakeProjectCreator/config/vc8_warnings.mpb: + Made vc8 and nmake more the same so that also with nmake the + crt warnings are suppressed + +Sun Mar 18 02:09:31 UTC 2007 Krishnakumar B <kitty@nospam.invalid.domain> + + * ace/OS_NS_stdlib.cpp: Fixed a race condition in + ACE_OS::mkstemp_emulation() where the value from + ACE_OS::gettimeofday() was truncated due to the use of + ACE_Time_Value::msec(void) as opposed to + ACE_Time_Value::msec(ACE_UINT64&) to convert it to milliseconds. + This resulted in seed value passed to ACE_OS::rand_r() returning + the same sequence of numbers when ACE_OS::mkstemp() was called + successively from the same process. As a result, the filenames + returned from ACE_OS::mkstemp() ended up being the same, if the + template passed in was the same between multiple successive + calls to ACE_OS::mkstemp(). + +Thu Mar 15 14:44:21 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * ace/Select_Reactor_Base.cpp: + The close() operation was trying to read from the notification + pipe event if the notification pipe was not open. I did not + notice this could happen, but the reactor constructor has a flag + for it. + +Thu Mar 15 12:06:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + Reverted the change below, the test stats show that after adding + the functions the Atomic_Op test does fail when these optimized + functions are enabled. + + Mon Mar 5 09:21:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + * ace/Atomic_Op_Sparc.c: + * ace/Atomic_Op.{h,cpp}: + Added optimized functions for SunCC on x86. Thanks to + Dennis Chernoivanov <cdi at tbricks dot com> for reporting this. + If you want to enable these, add atomic_ops_sparc=1 to your + platform_macros.GNU file + +Thu Mar 15 11:20:31 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * tests/Reactor_Notify_Test.cpp: + + Removed the auto_ptr that deleted the Reactor instance. It is + managed by the ACE library and doesn't need to be deleted. + +Wed Mar 14 21:34:03 UTC 2007 William R. Otte <wotte@dre.vanderbilt.edu> + + * bin/ciao_tests.lst + + Added requirement for the RT-CCM test called RT_CAPABLE, used to + indicate systems which are able to successfully run a test + involving RT policies. + +Wed Mar 14 12:32:36 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/tao_orb_tests.lst: + + Silly me, I forgot the script name when I added the + CORBA_e_compact_Implicit_Activation test. + +Wed Mar 14 09:09:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/config-hpux-11.00.h: + Set ACE_HAS_VSWPRINTF independent of the compiler used + +Wed Mar 14 08:09:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/config-linux-common.h: + * ace/config-lite.h: + * include/makeinclude/platform_linux_suncc.GNU: + Changes for the Sun Studio 12 Early Access compiler on linux + +Tue Mar 13 22:04:35 UTC 2007 Steve Huston <shuston@riverace.com> + + * tests/unload_libace.mpb: Squish the VC8 warning about deprecated + fopen(). + +Tue Mar 13 19:36:09 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/OS_NS_stdio.cpp (snprintf): Back off use of _vsnwprintf_s() in + the wide-char version as was done in the narrow-char version (see + Mon Mar 12 17:55:18 UTC 2007 Steve Huston <shuston@riverace.com>). + Also, in sprintf(), correct the selection of 3 vs 4 arg call for + WinCE. + +Tue Mar 13 18:43:28 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/tao_orb_tests.lst: + + Added a new test that only runs when the CORBA_E_COMPACT + configuration is enabled. + +Tue Mar 13 13:21:49 UTC 2007 Olli Savia <ops@iki.fi> + + * include/makeinclude/platform_lynxos.GNU: + Removed unnecessary PLATFORM_SSL_* settings. + +Tue Mar 13 12:13:37 UTC 2007 Steve Huston <shuston@riverace.com> + + * bin/MakeProjectCreator/config/vc8_warnings.mpb: For the + vc8_avoid_crt_secure_warnings feature, add /wd4996 to the nmake + case since it seems to not pay attention to _CRT_NONSTDC_NO_WARNINGS. + Also, reversed the default for VC8 SCL secure warnings - they're on + by default since Ossama implemented the checked STL iterator stuff. + +Tue Mar 13 09:57:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/tao_orb_tests.lst: + Added new DynUnion_Test + +Tue Mar 13 00:24:22 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * bin/MakeProjectCreator/config/vc8nmake.mpb: + + Subtract /EHsc from the compiler flags instead of /EHs. MPC now + uses /EHsc as the correct replacement for /GX. + +Mon Mar 12 23:44:25 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/checked_iterator.h (ACE_make_checked_array_iterator): + + Added documentation for this header. + + Changed length type from ptrdiff_t to size_t. The length refers + to the length of an array, not a pointer difference. + +Mon Mar 12 21:47:51 UTC 2007 Wallace Zhang <zhang_w@ociweb.com> + + * ace/OS_NS_stdio.inl: + + Fixed compilation errors on WinCE 5.0 builds. + +Mon Mar 12 22:26:08 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/checked_iterator.h (ACE_make_checked_array_iterator): + + New function template that returns a checked iterator if + supported by the platform (currently VC8 or better), and a + pointer otherwise. + + * ace/Array_Map.cpp (ACE_Array_Map, grow, operator==): + + Call new ACE_make_checked_array_iterator<>() iterator factory + function to generate destination iterator. Addresses VC8 + "C4996" warnings when security related warnings are enabled. + + * ace/Makefile.am: + * ace/ace.mpc: + * ace/ace_for_tao.mpc: + + Added new ace/checked_iterator.h header to the appropriate + header lists. + +Mon Mar 12 21:15:46 UTC 2007 Carlos O'Ryan <coryan@atdeesk.com> + + * tests/Reactor_Notify_Test.cpp: + The test was verifying the behavior described in bug 2845. I + neglected to fix the test when we fixed the bug. Mostly because + I am stupid, but also because the test only worked as expected + in a limited number of builds. + This change should fix the test problems we are seeing on the + scoreboard, but the only thing I can say for sure is: + + I managed to reproduce the crash on my machine. + + I understand why the test was crashing. + + I believe the test was wrong and not the library. + + I fixed the test and now valgrind is happy with it. + +Mon Mar 12 17:55:18 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/OS_NS_stdio.inl (vsnprintf): Back out the attempt to use the + new _vsnprintf_s() - it doesn't report the required string length + when it exceeds maxlen and we need this info. + +Mon Mar 12 17:38:26 UTC 2007 Steve Huston <shuston@riverace.com> + + * bin/MakeProjectCreator/config/vc8_warnings.mpb: + * bin/MakeProjectCreator/config/vc8nmake.mpb: + Moved the _CRT_NONSTDC_NO_WARNINGS, _CRT_SECURE_NO_DEPRECATE, and + _SCL_SECURE_NO_WARNINGS defs from vc8nmake.mpb to vc8_warnings.mpb + so they're all in one place for easy maintenance and also can be + switched on and off via MPC features. + + This effectively reverts the changes to vc8nmake.mpb since: + Fri Mar 9 21:04:22 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/config-win32-msvc-8.h: Deleted the pragma disable for warning + 4355 (this used in base member initializer list). Added it to + bin/MakeProjectCreator/config/vc8_warnings.mpb. This should both + unblock it for regular users and block it for the CIDLC build. + +Mon Mar 12 16:39:25 UTC 2007 Steve Huston <shuston@riverace.com> + + * ACE-INSTALL.html: Note that LabVIEW RT targets may require + the user to copy Windows DLLs to the target for Debug mode. + +Mon Mar 12 14:15:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/INET_Addr.cpp: + Removed not needed return + +Mon Mar 12 13:54:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/tao_other_tests.lst: + Don't run bug 2800 regression tests in a ST configuration + +Mon Mar 12 13:12:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/config-vxworks6.3.h: + * ace/config-vxworks6.4.h: + Only when building a static rtp we enable aio by default. For + building shared we need an optional library, see the file below + how to enable this + + * include/makeinclude/platform_vxworks6.3.GNU: + If you build rtp with shared library and want to use aio then you + need to build the custom AioPx library first, then add aio=1 + to your platform_macros.GNU file to enable aio in ACE and link + the needed additonal library + +Mon Mar 12 12:35:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/config-vxworks6.3.h: + * ace/config-vxworks6.4.h: + VxWorks 6.4 does deliver strings.h in kernel mode and strcasecmp/ + strncasecmp can't be used in rtp mode, but can be used in kernel + mode + +Mon Mar 12 10:51:38 UTC 2007 Steve Huston <shuston@riverace.com> + + * bin/MakeProjectCreator/config/vc8nmake.mpb: Added /D for + _CRT_SECURE_NO_DEPRECATE and _SCL_SECURE_NO_WARNINGS to silence + those compile warnings; added /wd4290 to silence the warnings + "C++ exception specification ignored except to indicate a function + is not __declspec(nothrow)". Note that the latter warning can also + be remedied by using ACE_THROW_SPEC; this has been done for ACE + but not in TAO. + +Mon Mar 12 01:31:16 UTC 2007 Abdullah Sowayan <abdullah.sowayan@lmco.com> + + * examples/DLL/test_dll.cpp: + + This example was not working under Windows XP when compiled with + Visual Studio 2003.NET. The culprit was the following line of code: + + dll.open (ACE_TEXT("./") ACE_DLL_PREFIX ACE_TEXT("DLL_Today")); + + We don't need to supply "./" to the open method, since $ACE_ROOT/lib + is in DLL search path environment variable (PATH for Windows and + LD_LIBRARY_PATH for unix-like systems). + +Fri Mar 9 21:04:22 UTC 2007 Steve Huston <shuston@riverace.com> + + * bin/MakeProjectCreator/config/vc8nmake.mpb: Added /wd4250 + (to disable inheritance via dominance warning) and + /D _CRT_SECURE_NO_DEPRECATE to disable the warnings about + deprecated CRT functions. + +Fri Mar 9 16:49:03 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/OS_NS_string.inl (strncat): Disable the TR24731/2005 calls to + strncat_s() and wcsncat_s() for now. The dest length includes the + characters that were already in dest before appending src. To really + use the safer functions, the dest max length is needed from the + caller. So, we'll either need a new method or an added arg to the + current methods. + +Fri Mar 9 16:27:23 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/OS_NS_stdio.cpp (snprintf): + * ace/OS_NS_string.inl (strncat): When using the "safe" calls from + TR24731/2005, specify the dest count as the user-specified source + length + 1 for the null. This assumes the caller knows the dest + buffer is long enough to hold what's specified. Specifying + _TRUNCATE as the src count gets the behavior we expect - the + specified number of bytes are copied with the assurance a nul + terminator is added. This avoids the CRT error handler. + +Fri Mar 9 16:10:09 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/Log_Msg_NT_Event_Log.{h cpp}: Changed return type of log() from + int to ssize_t. Matches changes from: + Fri Mar 9 00:53:43 UTC 2007 Steve Huston <shuston@riverace.com> + +Thu Mar 9 09:44:14 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Time_Value.cpp (operator *=): + + Reverted bulk of the changes. New implementation will be + revisited as time permits. + +Fri Mar 9 09:42:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/platform_vxworks6.3.GNU: + No need to explicitly link dl when building shared + +Thu Mar 9 07:44:14 UTC 2007 Ossama Othman <ossama_othman@symantec.com> + + * tests/Time_Value_Test.cpp (run_main): + + Simplified assertions. Rather than test equality of sec() and + usec() components separately, just test equality of the + ACE_Time_Value objects as a whole. + +Fri Mar 9 00:53:43 UTC 2007 Steve Huston <shuston@riverace.com> + + Changes to resolve or quiet warnings from Visual Studio .NET 2005: + + * bin/MakeProjectCreator/config/acedefaults.mpb: + * bin/MakeProjectCreator/config/vc8_warnings.mpb: Removed the + default addition of _CRT_SECURE_NO_DEPRECATE from the MPC VC8 + project templates. Many of the places where the changes recommended + by Microsoft could be made have been made (see further below). There + are still a number of places where the warnings are generated, + mostly because ACE wraps "unsafe" methods such as strcpy(), printf(), + etc. These can't legitimately be changed. So most of what can (and + should) be changed in ACE has been. To keep the warnings quiet + without removing them from any user application that includes ACE + headers, the acedefaults.mpb base project now inherits from + vc8_warnings - vc8_warnings will, by default, silence the warnings + we don't really need to see, but users of ACE will still get the + warnings when building their apps, unless they take similar steps + to silence them (or if they generate projects using the acedefaults + base project with MPC). + We can still elect to enable all the warnings by setting MPC features + to 0 (see vc8_warnings.mpb for features). And this is a good thing + to do still since we can improve the security situation by reviewing + things flagged as warnings and repair what's needed (not everything + flagged will be in need of a fix). + + * ace/config-win32-msvc-8.h: Removed most of the pragmas that silenced + legit warnings from VC8. Added ACE_HAS_TR24731_2005_CRT indicating + that there's a C Run-Time implementing C99 draft TR 24731 dated + October 2005 detailing security improvements to the C run-time. + This macros indicates that the more secure versions of the calls + can be used where desireable. + Also added a macro, ACE_WIN32_VC8, to easily tell where we need to + selectively disable a warning that we can tell by inspection is + irrelevant. Such as use of 'this' in an initializer list and we can + tell by inspecting the code in question to be sure that the passed + pointer isn't used to call a method on an object that's not fully + constructed yet. + + * ace/config-macros.h: When ACE_HAS_TR24731_2005_CRT is defined, + define ACE_SECURECRTCALL to make a call and adjust the returned + error value to the ACE return value convention. + + * ace/README: Describe ACE_HAS_TR24731_2005_CRT. + + * NEWS: Described the removing of the warning preventers and that + user code may trigger warnings that weren't showing up in previous + ACE versions. + + * ace/Global_Macros.h: For VC8 with ACE_HAS_EXCEPTIONS, define + ACE_THROW_SPEC as throw(...) always. VC8 always treats any throw + spec as throw(...) and issues a warning to this effect. This change + silences the warnings. The #if intends to set this for VC8 only, + guessing that any VC9 will be _MSC_VER 1500. + + * ace/os_includes/sys/os_types.h: For non-Borland, non-Mingw ACE_WIN32 + change mode_t from u_short to int. Matches CRT docs. + + * ace/Local_Tokens.cpp: + * ace/Select_Reactor_Base.inl: + * ace/WFMO_Reactor.cpp: + * apps/Gateway/Gateway/Event_Channel.cpp: + Disable VC8's C4355 warning that 'this' is used in an initializer + list; we know, and it's ok. + + * ace/Select_Reactor_T.cpp (ctor): Rather than pass a reference to + 'this' incompletely constructed object to token_, call token_'s + reactor() method to set the reactor impl pointer. + + * ace/Registry.{h cpp}: Rearranged setting of Binding_Iterator in the + inherited Iteration_State object. Instead of passing a reference to + an object that's not yet fully constructed, call a new iterator() + method to set that pointer. + + * ace/Service_Config.cpp (ctor): Rather than set the TSS pointer in + the initializer list before it's complete, call ts_object() to set + it within the constructor body. We could also catch a setup error + here and throw an exception if ACE ever goes that way. + + * ace/OS_NS_time.h (ace_timezone): Use the GetTimeZoneInformation() + system call to get timezone bias for all Windows platforms and + compilers. Much easier and faster than going through the C run-time + and will work with new US daylight savings changes automatically. + + * ace/OS_NS_sys_stat.inl: (umask): + * ace/OS_NS_stdio.inl (vsnprintf): + * ace/OS_NS_stdio.cpp (sprintf, snprintf): + * ace/OS_NS_string.inl (strtok_r, strncat): + * ace/OS_NS_string.cpp (strerror): + Use "safer" functions when ACE_HAS_TR24731_2005_CRT. + + * ace/OS_NS_time.{inl cpp}: (asctime_r, ctime_r, gmtime_r, + localtime_r): Use new functions when ACE_HAS_TR24731_2005_CRT. + NOTE - I purposely did not substitute the secure version for the + non _r methods (asctime, ctime, etc.) since that would have to use + a local static variable. That defeats the purpose of being able to + use a caller-supplied destination. Thus, the places that can + substitute a secure function are, and those that can't don't. This + will notify users of code that's potentially unsafe and allow them to + switch to the safer _r version. We can set the + _CRT_SECURE_NO_WARNINGS macro for non-inline builds (where the unsafe + versions will be compiled in) and leave the warning in place for + inlined builds to put the user on notice that a potentially unsafe + method is being used. + + * ace/UUID.cpp (from_string_i): Use sscanf_s for VC8. + + * ace/Atomic_Op_T.{cpp inl}: + * ace/INET_Addr.cpp: + * ace/Filecache.cpp: + * ace/Future.cpp: + * ace/Map_Manager.inl: + * ace/Message_Queue_T.cpp: + * ace/Process_Mutex.cpp: + * ace/RW_Process_Mutex.cpp: + * ace/Stream.cpp: Don't use 'this' in initializer list. Quiets + VC8's warnings. + + * protocols/ace/TMCast/Group.{hpp cpp}: Change 'throw' to + ACE_THROW_SPEC to quiet the VC8 warnings. + + * tests/unload_libace.mpb: Add _CRT_SECURE_NO_WARNINGS for VC8. Else + VC8 warns about fopen() and we don't care. + + * ace/README: Added ACE_LACKS_WIN32_GETPROCESSTIMES. + + * ace/config-WinCE.h: Added ACE_LACKS_WIN32_SECURITY_DESCRIPTORS and + ACE_LACKS_WIN32_GETPROCESSTIMES. + + * ace/OS_NS_dirent.cpp (opendir_emulation): WinCE offers + GetFileAttributes() but documents a failing return value as + 0xFFFFFFFF whereas on desktop/server Windows a failed call returns + INVALID_FILE_ATTRIBUTES. So on WinCE, define INVALID_FILE_ATTRIBUTES. + + * ace/OS_NS_sys_resource.inl (getrusage): Don't try GetProcessTimes() + on Win32 if ACE_LACKS_WIN32_GETPROCESSTIMES is set. + + * ace/OS_NS_stdio.inl (cuserid): WinCE returns ENOTSUP for these. + + * ace/OS_NS_unistd.inl (chdir): If ACE_HAS_WINCE, return ENOTSUP. + + * ace/Log_Record.cpp (operator<<): Truncate the message data length + down to 32 bits (ACE_CDR::ULong) size since that's all the + CDR ops and header field can handle. + + * ace/Log_Msg_Backend.h: + * ace/Log_msg_UNIX_Syslog.{h cpp}: + * ace/Log_Msg_IPC.h: + * ace/Log_Msg_IPC.cpp (log): Changed the return type of log() from int + to ssize_t to match what it's returning (byte count or error). + Corrected the max_payload_size calculation + to match the header that operator<< (above) writes. The MAXLOGMSGLEN + isn't always enough now that ACE_Log_Record's data can be extended. + +Fri Mar 9 00:57:26 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * ace/Event_Handler.cpp: + Fixed bug 2845, the event handler destructor no longer calls + purge_pending_notifications. More details about the rationale + can be found in the bug report. + + * NEWS: + * docs/ACE-FMM.html: + Prominently document the change, since it may catch people + unawares. + +Thu Mar 8 21:27:54 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/String_Base.{h cpp}: Change ACE_String_Base::release_ from int + to bool. Thanks to Kelly Hickel for this change. This changes the + signatures of the following methods: + ACE_String_Base (const CHAR *s, + ACE_Allocator *the_allocator = 0, + int release = 1); + to + ACE_String_Base (const CHAR *s, + ACE_Allocator *the_allocator = 0, + bool release = true); + + ACE_String_Base (const CHAR *s, + size_type len, + ACE_Allocator *the_allocator = 0, + int release = 1); + to + ACE_String_Base (const CHAR *s, + size_type len, + ACE_Allocator *the_allocator = 0, + bool release = true); + + void set (const CHAR * s, int release = 1); + to + void set (const CHAR * s, bool release = true); + + void set (const CHAR * s, size_type len, int release); + to + void set (const CHAR * s, size_type len, bool release); + + void clear (int release = 0); + to + void clear (bool release = false); + + * ace/Configuration.cpp: + * ACEXML/common/NamespaceSupport.cpp: + * ACEXML/parser/parser/Entity_Manager.inl: + * tests/SString_Test.cpp: Changes to string usage to + reflect the above signature changes. Changed '0' args to 'false'. + + * NEWS: Added a description of the above changed methods. + +Thu Mar 8 21:24:38 UTC 2007 Steve Huston <shuston@riverace.com> + + * tests/Cached_Conn_Test.cpp: + * tests/Cached_Accept_Conn_Test.cpp: Replace use of ACE_ASSERT with + ACE_ERROR. ACE_ASSERTS on single-process targets do really bad + things... + +Thu Mar 8 20:29:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/platform_vxworks6.3.GNU: + More changes to build with shared library support using VxWorks 6.x + +Thu Mar 8 19:06:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/platform_vxworks6.3.GNU: + Only add dl to LDLIBS when building shared + +Thu Mar 8 18:06:14 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Time_Value.cpp (operator *=): + + Removed vestigial time value saturation code that does not work + correctly with latest implementation. + +Thu Mar 8 14:21:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/config-vxworks6.2.h: + * ace/config-vxworks6.3.h: + * ace/config-vxworks6.4.h: + Don't set ACE_HAS_SVR4_DYNAMIC_LINKING when we do a static + rtp build. + +Thu Mar 8 13:04:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/platform_vxworks6.3.GNU: + Added some initial support for building shared libraries + with vxworks 6.3, this is not ready yet, in progress + +Thu Mar 8 12:17:50 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/MakeProjectCreator/config/ace_openssl.mpb: + * include/makeinclude/wrapper_macros.GNU: + + Moved the addition of SSL related macros out of wrapper_macros.GNU + and into a specific section in ace_openssl.mpb. The result is + that the SSL related information is only added to the projects + that inherit from ace_openssl (instead of every single project). + +Thu Mar 8 02:59:02 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * ace/Notification_Queue.cpp: + Fixed implementation of reset() make sure the code works right + when there are null event handlers in the queue. This should + have been part of the original fixes for bug 2820. + + * tests/Notification_Queue_Unit_Test.cpp: + Add new test cases for the reset() operation, in particular + check for null event handlers in the queue. + +Wed Mar 7 19:07:58 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/Message_Block.cpp: + + Fixing a problem caused by inadvertently overwriting an external + buffer passed in the constructor. It happens when + ACE_INITIALIZE_MEMORY_BEFORE_USE is set (as result of setting + ACE_HAS_VALGRIND). Thanks to Johnny Willemsen for reporting it. + +Wed Mar 7 18:44:10 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * include/makeinclude/wrapper_macros.GNU: + + Fixed an issue where if ssl wasn't set to zero, it was assumed to + be set to non-zero (ignoring the case where it could be set to + empty). + +Wed Mar 7 17:32:51 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Time_Value.cpp (operator*=): + + Altered implementation to avoid converting the microseconds + component of the ACE_Time_Value to a fraction of a second. In + the case were the floating point type's mantissa width is the + same as the width of the time_t (seconds) value, a maxed out + ACE_Time_Value seconds component would utilize all bits in the + mantissa, leaving no room for a fraction. The microseconds + component would end up not contributing to the multiplication. + Multiply the seconds and microseconds components separately + instead. This also saves us a floating point division, which + can be expensive. + +Wed Mar 7 16:35:04 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/MakeProjectCreator/config/ace_openssl.mpb: + * include/makeinclude/wrapper_macros.GNU: + + Removed the addition of $(SSL_ROOT)/include and $(SSL_ROOT)/lib + from the includes and libpaths project variables and ssl and + crypto from the lit_libs settings for gnuace (as was done for + automake). + + This functionality will now come from wrapper_macros.GNU in the + form of PLATFORM_SSL_CPPFLAGS, PLATFORM_SSL_LDFLAGS and + PLATFORM_SSL_LIBS. If these variables are not set, they will be + given defaults based on the SSL_ROOT variable, as long as SSL_ROOT + is set and is not set to /usr. + + * bin/MakeProjectCreator/templates/gnu.mpd: + + Added support for the compile_flags and linkflags template + variables. + +Wed Mar 7 13:12:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/platform_vxworks5.5.x.GNU: + * include/makeinclude/platform_vxworks6.2.GNU: + Simplified these files + + * include/makeinclude/platform_vxworks6.3.GNU: + Added support to use a windows host system for the cross build + +Wed Mar 7 12:34:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/wrapper_macros.GNU: + Removed setting of ACELIB, in the generated makefiles it is + always cleared again + + * bin/MakeProjectCreator/templates.gnu.mpb: + Removed clearing of ACELIB and setting of tao_dont_use_idl_make_rule + +Wed Mar 7 12:20:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/rules.local.GNU: + Removed check for TAO_IDL_DEP, the generated GNUmakefiles do + check for this + +Wed Mar 7 11:46:46 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * tests/Bug_2820_Regression_Test.cpp: + Use the standard ACE workaround for platforms with a broken + std::auto_ptr<>. + +Wed Mar 7 09:56:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Atomic_Op.cpp: + Implemented the single_cpu_* methods for PPC, because of the lack + of a multi cpu ppc board we don't have these methods implemented + and we don't enable this by default. If you want to use the PPC + optimized operations for a single cpu board add + ACE_HAS_BUILTIN_ATOMIC_OP to your config.h + +Wed Mar 7 02:51:50 UTC 2007 William R. Otte <wotte@dre.vanderbilt.edu> + + * docs/bczar/bczar.html + + Updated bczar documentation to reflect change in release script. + + * bin/make_release + * bin/make_release.py + + Removed the old release script and added the new one. + +Tue Mar 6 23:06:15 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Time_Value.cpp (operator*=): + + Use long double for internal computations if double is not large + enough to retain precision of the time_t/suseconds_t values. + This will only provide improved behavior with compilers that + support double-extended precision floating numbers, i.e. >= 79 + bits. + + Added comments to explain what is being done in the + implementation. + +Tue Mar 6 07:47:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/OS.inl: + Fixed compile error with VxWorks 6.x using a Windows host + +Tue Mar 6 15:27:52 UTC 2007 William R. Otte <wotte@dre.vanderbilt.edu> + + * NEWS: + + Noted removal of eVC3/4 projects. + +Tue Mar 6 15:00:00 UTC 2007 Ciju John <johnc@ociweb.com> + + * tests/INET_Addr_Test.cpp: + Add test for ACE_INET_Addr::is_loopback () API. This tests a + selected set of addresses to validate API's conformance with RFC + 3330. + +Tue Mar 6 13:21:24 UTC 2007 Douglas C. Schmidt <schmidt@dre.vanderbilt.edu> + + * COPYING: Updated this file to clarify that the ACE license need + not be shipped with binary-only releases. Thanks to Michael + Kircher <michael dot kircher at siemens dot com> and Werner + Burger <werner dot burger at siemens dot com> for motivating + this. + +Tue Mar 6 12:43:01 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * tests/Date_Time_Test.cpp: + + Handle the rare case where the minute rolls over during the test. + +Tue Mar 6 04:41:36 UTC 2007 Abdullah Sowayan <abdullah.sowayan@lmco.com> + + * examples/DLL/README: + + README was referring to files (dsp/dsw) used by Visual Studio C++ 6. + Updated to refer to files (sln/vcproj) used by VC++ 7.1 or later. + +Tue Mar 6 02:37:53 UTC 2007 Carlos O'Ryan <coryan@atdeesk.com> + + * Merged changes from the fix_bug_2820 branch. From revision + 77533 to revision 77546. + + Tue Mar 6 02:07:43 UTC 2007 Carlos O'Ryan <coryan@atdeesk.com> + + * ace/Select_Reactor_Base.cpp: + Fixed bugs in loop to extract event handlers from notification + pipe. + + * tests/Bug_2820_Regression_Test.cpp: + Fixed several problems in the test: + - The Event_Handler called on the Reactor on its destructor, but + by then the reactor was already destroyed. + - Improved debugging messages. + - Fixed memory management problems, valgrind is happy now. + + Mon Mar 5 18:01:36 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * ace/Notification_Queue.cpp: + When deleting the queue we must decrease the reference count on + all the event handlers stored there. + + * ace/Select_Reactor_Base.cpp: + When the notification pipe is destroyed we must read the data + stored there and release all the event handlers. Otherwise the + reference counts are broken. + + * tests/run_test.lst: + Add Bug_2820_Regression_Test to the list. + + Mon Mar 5 17:16:16 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * tests/Bug_2820_Regression_Test.cpp: + The test was broken, it did not delete the Reactor, so it failed + to test what we wanted. + +Tue Mar 6 00:10:36 UTC 2007 Steve Huston <shuston@riverace.com> + + * examples/ASX/UPIPE_Event_Server/event_server.cpp: Use ACE_DEBUG + to display the message containing a time_t. Use of the standard + iostream operator<< with time_t gives the Intel90 compiler fits. + +Mon Mar 5 23:09:59 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/Log_Msg.{h cpp}: Added new format specifiers: + %b to format ssize_t values + %B to format size_t values + %: to format time_t values + + Note that these use, when necessary, printf format specifiers such + as ACE_SIZE_T_FORMAT_SPECIFIER and ACE_INT64_FORMAT_SPECIFIER. + The printf formatters can't be used portably in ACE_Log_Msg format + strings because 1) some of them use %I, which has a different meaning + in ACE_Log_Msg already and 2) some of the new printf formats are + more complicated that ACE_Log_Msg knows how to deal with in its + existing logic to use format modifiers. The new ACE_Log_Msg format + specifiers make it easy to portably log these types that can change + sizes across platforms. + + * ace/Log_Record.cpp (dump): + * tests/Cached_Allocator_Test.cpp: + * tests/High_Res_Timer_Test.cpp: Use the new ACE_Log_Msg format + specifiers for size_t and time_t logged values. + +Mon Mar 5 22:23:01 UTC 2007 Adam Mitz <mitza@ociweb.com> + + * bin/clean_sems.sh: + + s/xargs/xargs -r/ The -r switch to xargs prevents the command from + being executed when there are no arguments. This will prevent the + extra scoreboard output on Linux (when rem_sems is used): + usage: ipcrm [ [-q msqid] [-m shmid] [-s semid] + [-Q msgkey] [-M shmkey] [-S semkey] ... ] + deprecated usage: ipcrm {shm | msg | sem} id ... + +Mon Mar 5 18:02:42 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Basic_Types.h (ACE_UINT64_MAX): + + Simplified preprocessor conditionals. + + * ace/RB_Tree.h (ACE_RB_Tree_Iterator_Base): + + Made copy constructor public instead of private. The assignment + operator was already public so there was nothing to be gained by + making the copy constructor protected. Addresses Sun Studio 10 + warnings. + + * ace/Time_Value.inl (to_usec): + + Added support for ACE_LACKS_UNSIGNEDLONGLONG_T case. + + * ace/UUID.cpp (get_systemtime): + + In the ACE_LACKS_UNSIGNEDLONGLONG_T case, wrap 64-bit integer + constant with ACE_INT64_LITERAL macro. Addresses potential + problem where the constant is intepreted as a "long", instead of + "long long". + +Mon Mar 5 17:22:26 UTC 2007 Ciju John <johnc@ociweb.com> + + * ace/INET_Addr.inl: + * ace/INET_Addr.cpp: + * ace/QoS/SOCK_Dgram_Mcast_QoS.cpp: + RFC 3330 defines loopback as any address with 127.x.x.x. The + is_loopback API currently simply compares against the + INADDR_LOOPBACK macro. Doing so restricts loopback definition to + 127.0.0.1 which is inconsistent with the RFC definition. This + change will include any address with 127.x.x.x as loopback. + +Mon Mar 5 14:16:49 UTC 2007 Olli Savia <ops@iki.fi> + + * tests/run_test.lst: + Updated for LynxOS. + +Mon Mar 5 11:25:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/config-hpux-11.00.h: + Enabled ACE_HAS_VSFWPRINTF for aCC compiler + +Mon Mar 5 09:21:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Atomic_Op_Sparc.c: + * ace/Atomic_Op.{h,cpp}: + Added optimized functions for SunCC on x86. Thanks to + Dennis Chernoivanov <cdi at tbricks dot com> for reporting this. + If you want to enable these, add atomic_ops_sparc=1 to your + platform_macros.GNU file + +Sat Mar 3 17:35:52 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Basic_Types.h (ACE_UINT64_MAX): + + Re-wrap the integer constant with the ACE_UINT64_LITERAL macro + to ensure the appropriate type specifier is used. + +Fri Mar 3 5:01:19 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Basic_Types.h (ACE_UINT64_MAX): + + Reintroduced this constant for the case where 64 bit integers + are natively supported. It was inadvertently removed. + + Removed stray semi-colon in ACE_LACKS_{UNSIGNED}LONGLONG_T + case. + +Fri Mar 2 21:42:54 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Basic_Types.h (ACE_UINT64_MAX): + + Corrected value for the ACE_LACKS_{UNSIGNED}LONG_LONG_T case. + + (ACE_U64_TO_U32, ACE_CU64_TO_CU32): + + Converted these macros to inlined functions. + + (ACE_LONGLONG_TO_PTR): + + Cast pointer to intptr_t instead of ptrdiff_t before casting + to the pointer type. The ptrdiff_t type is meant for pointer + differences. + + * ace/High_Res_Timer.cpp (dump): + + Drop ACE_CU64_TO_CU32 macro in favor of standard static_cast<>. + There was no real use for that macro in this case. + + * ace/Numeric_Limits.h (ACE_Numeric_Limits): + + Added a specialization for ACE_U_LongLong. + + * ace/Sample_History.cpp (dump_samples): + + Removed unnecessary cast to 32 bit value. + + Corrected Log_Msg format specifier for unsigned type. + + * ace/Time_Value.cpp (operator*=): + * tests/Time_Value_Test.cpp (run_main): + + Removed ACE_LACKS_NUMERIC_LIMITS preprocessor conditional. Use + ACE_Numeric_Limits<> instead. It is always available. + + * ace/Time_Value.h (to_usec): + * ace/Time_Value.inl (to_usec): + + Enabled support for this method when ACE_LACKS_LONGLONG_T is + defined. + + * ace/UUID.cpp (get_systemtime): + + Fixed support for ACE_U_LongLong type. + +Fri Mar 2 20:48:45 UTC 2007 Steve Huston <shuston@riverace.com> + + * ACE-INSTALL.html: Improved the description of using Pharlap ETS and + described the ACE_PHARLAP_TESTLOG_TO_FILE config setting. + + * ace/config-pharlap.h: Add a IN_CLASSD(i) macro that always reports + false. Although Pharlap is supposed to support multicast, the + IN_CLASSD macro is only added at WinSock 2. + + * tests/Test_Output.cpp: Allow Pharlap to log to a file by setting + ACE_PHARLAP_TESTLOG_TO_FILE in config.h. The default behavior of + logging to the system console still stands. + +Fri Mar 2 18:18:25 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/MakeProjectCreator/modules/GNUACEWorkspaceCreator.pm: + + Override the targets method to supply a list of targets that can + be used with Eclipse. + +Fri Mar 2 17:48:09 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Truncate.h (Truncator): + + Fixed one more typo in the ACE_U_LongLong partial specialization + of this functor. Also corrected truncation logic. + +Thu Mar 1 16:30:13 UTC 2007 Douglas C. Schmidt <schmidt@dre.vanderbilt.edu> + + * ace/Log_Msg.h (ACE_Log_Msg): Updated the document to remove the + claim that stdout can be used. Thanks to Ian C White + <Ian_C_White at raytheon dot com> for reporting this. + +Thu Mar 1 13:15:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/config-hpux-11.00.h: + Removed ACE_HAS_VSWPRINTF, there are dependencies on this + define and no time to handle this fully. + +Thu Mar 1 12:28:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Sock_Connect.cpp: + Fixed sun warning + +Thu Mar 1 12:18:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * examples/Reactor/Misc/test_signals_2.cpp: + Removed HPUX disabling, compiles fine on our system + +Thu Mar 1 9:42:12 UTC 2007 Martin Corino <mcorino@remedy.nl> + + * bin/PerlACE/Process_Unix.pm: + * bin/PerlACE/ProcessVX_Unix.pm: + Some fixes for running VxWorks 6.x RTP tests. + +Thu Mar 1 08:51:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/OS_NS_stdio.{cpp,inl}: + Introduced new ACE_HAS_VSWPRINTF that we can set when a platform + has this method + + * ace/config-hpux-11.00.h: + Added ACE_HAS_VSWPRINTF + +Wed Feb 28 21:34:45 UTC 2007 Wallace Zhang <zhang_w@ociweb.com> + + * bin/tao_orb_tests.lst: + + Added two new TAO tests. + +Wed Feb 28 19:16:12 UTC 2007 Martin Corino <mcorino@remedy.nl> + + * ace/DLL_Manager.cpp: + * ace/Parse_Node.cpp: + Fixed pointer conversion warnings on OpenVMS. + +Wed Feb 28 17:44:20 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Truncate.h (Truncator): + + Fixed typo in the ACE_LACKS_UNSIGNEDLONGLONG_T case. + s/ACE_UIN32/ACE_UINT32/. + +Wed Feb 28 13:32:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/config-hpux-11.00.h: + Moved ACE_HAS_SSIZE_T outside of the compiler specific part, gcc + and aCC do have ssize_t. Also moved ACE_HAS_XPG4_MULTIBYTE_CHAR + outside of the compiler part. + +Wed Feb 28 11:09:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/OS_NS_unistd.cpp (num_processors): + Implemented for HPUX + +Wed Feb 28 09:58:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/ADDIDL.DSM: + Moved to TAO + + * bin/msvc_cidlc.pl: + Moved to CIAO + + * indent_macros.perl: + * indent_macros.pl: + Renamed extension to .pl + +Wed Feb 28 09:44:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/corba-boolean.perl: + * bin/rm_exception_macros.pl: + * bin/subst_env.pl: + * bin/tao_env.pl: + * bin/tao_svcconf.pl: + Moved to TAO/bin, no need to ship these with ACE + +Wed Feb 28 08:14:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/DLL_Manager.cpp: + Fixed a deadlock that could occur when a dll is loaded which itselfs + also loads a dll in its static initialisation. This fixes bugzilla bug + 2830. + +Tue Feb 27 21:18:07 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * netsvcs/lib/Time_Request_Reply.cpp: + * tests/Integer_Truncate_Test.cpp: + * tests/CDR_File_Test.cpp: + + s/ACE_Utils::Truncate/ACE_Utils::truncate_cast/g. The former is + deprecated. + + * tests/High_Res_Timer_Test.cpp: + + Improved variable types to make use of + ACE_Utils::truncate_cast<> unnecessary. + +Tue Feb 27 18:55:12 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/CDR_Stream.cpp: + * ace/Message_Block.cpp: + + The intent of the change is to better detect (and be able to + recover from) a situation when memory allocation fails in + constructing or initializing an ACE_Data_Block. It fixes a + problem with the change from "Mon Feb 26 19:06:36 UTC 2007" + which wasn't accounting for the new datablock size correctly. + +Tue Feb 27 19:33:13 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Truncate.h: + + Added specializations for ACE_U_LongLong. + + Added yet another const specialization for BCB6 to work around + its inability to use valid specializations with template + parameters that are not const. + + Removed ACE_Utils::Truncate<> altogether. Use + ACE_Utils::truncate_cast<> instead. + +Tue Feb 27 15:18:50 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/MakeProjectCreator/docs/templates/gnu.txt: + + Added documentation for template variables used by the gnuace + project type. This is used by document_template.pl found in + MPC/devtools. + +Tue Feb 27 14:59:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * tests/Process_Manager_Test.cpp: + Fixed incorrect initialized variable + +Tue Feb 27 14:20:26 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/CDR_Stream.cpp: + * ace/Message_Block.cpp: + + Reverting my patch from "Mon Feb 26 19:06:36 UTC 2007". It needs + more work. + +Tue Feb 27 13:57:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/High_Res_Timer.cpp: + * ace/DLL_Test.cpp: + Removed check for KCC, that compiler is not supported anymore + +Tue Feb 27 11:45:06 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * tests/Notification_Queue_Unit_Test.cpp: + Fixed compilation error in Redhat_Enterprise_Linux_3_Debug_WChar + build. Thanks to Johnny Willemsen for letting me know. + +Tue Feb 27 03:34:40 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * tests/Notification_Queue_Unit_Test.cpp: + Use functions instead of macros to workaround compiler warnings + in the HP-UX_11iv2_IA64_aCC_wchar build. + The warnings were harmless, but the code is more readable this + way in any case. Thanks to Johnny Willemsen for bringing the + warnings to my attention. + +Mon Feb 26 20:58:05 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/Service_Gestalt.cpp (insert): + + Removed the usage of fprintf to print messages about service + descriptors, inserted using static initializers. It is the only + way if you want to see what is being inserted before ACE_Log_Msg + gets initialized but because it uses stderr directly, it may + affect some aplications' expected output. + + * ace/Service_Repository.cpp (insert): + + Fixing a defective debug statement. + +Mon Feb 26 19:06:36 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/CDR_Stream.cpp (grow_and_adjust, clone_nocopy): + + For large sizes the message block constructor may fail to + allocate and/or initialize the necessary ACE_Data_Block + instances. The failure to initialize can go unnoticed and may + cause an unlimited number of recursive invocations of + grow_and_adjust. The change adds a test for the invariant of the + newly allocated ACE_Message_Block instance. + + * ace/Message_Block.cpp (clone_nocopy): + + When memory is scarce the allocation of ACE_Data_Block may + succeed while the initialization can fail. Adding a test, + before the message block is declared good. + +Mon Feb 26 17:17:08 UTC 2007 Adam Mitz <mitza@ociweb.com> + + * bin/PerlACE/Run_Test.pm: + + Set up the $PerlACE::CIAO_ROOT variable, just like $PerlACE::TAO_ROOT + except for CIAO. + + * bin/auto_run_tests.pl: + + Find CIAO tests relative to the $CIAO_ROOT directory. + +Mon Feb 26 13:56:42 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/ACETAOCIAO.style: + Updated to version 1.12 of StyleManager + +Mon Feb 26 13:03:19 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/MakeProjectCreator/templates/gnu.mpd: + + Added code to surround all libraries with a -Wl,--start-group and + -Wl,--end-group only if the compiler is g++, we're linking static + libraries, it's an executable and the 'link_groups' template + variable is set. + +Mon Feb 26 10:08:19 UTC 2007 Olli Savia <ops@iki.fi> + + * tests/SOCK_Dgram_Bcast_Test.cpp: + Pass correct buffer length to ACE_OS::vsnprintf(). + Thanks to Yves Alloyer <alloyer at ipanematech dot com> for + reporting this and providing initial patch. + +Mon Feb 26 08:50:13 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/Process.cpp (command_line): + Pass correct buffer length to ACE_OS::vsnprintf(). + Thanks to Yves Alloyer <alloyer at ipanematech dot com> for + reporting this and providing initial patch. + +Sun Feb 25 17:26:37 UTC 2007 Jaiganesh B <jai.dre.vanderbilt.edu> + + * bin/tao_orb_tests.lst: + + Added the DiffServ policy test's run_test.pl. + +Sat Feb 24 23:01:16 UTC 2007 Steve Huston <shuston@riverace.com> + + * configure.ac: Added ACE_CHECK_LACKS_FUNCS(alphasort) + +Sat Feb 24 22:51:09 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/config-sunos5.5.h: + * ace/config-sunos5.10.h: Add ACE_LACKS_ALPHASORT at Solaris 2.5 and + remove it at Solaris 10. Solaris 8 and 9 have an alphasort() but + only for use with the BSD cc compiler and with disclaimer that it's + not supported for use with system libraries nor with multiple + threads. Solaris 10 brings it into supported use. + +Sat Feb 24 17:03:47 UTC 2007 Steve Huston <shuston@riverace.com> + + * examples/NT_Service/main.cpp: + * examples/NT_Service/nt_svc.{h cpp}: Don't check for disabling + macros until a config file has been included, else it never builds + anywhere. Thanks to Johnny Willemsen for pointing this out. + +Sat Feb 24 09:01:19 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/config-linux.h: + + AIO support pulls in the rt library, which pulls in the pthread + library. Disable AIO in single-threaded builds. + + * tests/Recursive_Mutex_Test.cpp: + + Check if ACE_Recursive_Thread_Mutex::get_nesting_level() is + supported before attempting to use it. + +Sat Feb 24 08:34:08 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/config-openvms.h: + * ace/config-qnx-rtp-62x.h: + * ace/config-qnx-rtp-pre62x.h: + Added ACE_LACKS_ALPHASORT. + +Fri Feb 23 23:10:17 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * tests/Recursive_Mutex_Test.cpp (test_recursion_depth): + + Issue error messages instead of assert()ing. VxWorks builds + will hang on a crash. Thanks to Johnny for pointing this out. + +Fri Feb 23 19:24:30 USMST 2007 Yan Dai <dai_y@ociweb.com> + + * performance-tests/SCTP/SOCK_SEQPACK_srv.cpp: + + Made run_server() and other functions and structures only compiled + when ACE_HAS_SCTP is defined. This should resolve the warning that the + run_server() is defined but not used on some platforms. + +Fri Feb 23 17:47:28 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * tests/Bug_2815_Regression_Test.cpp: + The test should be enabled only if the notification queue is + also enabled. Basically purge_pending_notifications() is not + supported without the notification queue, and that is the + operation we want to test. + +Fri Feb 23 15:37:03 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/OS_NS_dirent.inl (alphasort): Use a bigger hammer on the casts. + +Fri Feb 23 10:50:41 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/OS_NS_dirent.inl: + Fixed casts again. Hopefully now I got it right. + +Fri Feb 23 08:56:15 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/OS_NS_dirent.inl: + Added casts to fix compile errors. + + * ace/config-rtems.h: + * ace/config-vxworks5.x.h: + * ace/config-vxworks6.2.h: + * ace/config-vxworks6.3.h: + * ace/config-vxworks6.4.h: + Added ACE_LACKS_ALPHASORT. + +Thu Feb 22 18:37:00 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/Service_Config.h: + * ace/Service_Config.inl: + * ace/Service_Config.cpp: + * ace/Service_Gestalt.h: + * ace/Service_Gestalt.cpp: + + Fixing build troubles caused by the different ACE_TSS interface + for single-threaded builds. Eliminated the intermediary class + (TSS_Resources) and now storing the actual service gestalt + pointer in TSS, thus reducig the memory allocation overhead. + +Thu Feb 22 16:45:30 UTC 2007 Steve Huston <shuston@riverace.com> + + * tests/MT_SOCK_Test.cpp: Document the actual cause of what's been + observed for some time with some Windows systems exhibiting a + "delayed rejection" of connections in excess of the listener's + backlog. Thank you to Aleksandar Vukajlovic <vukajlo at finsoft + dot co dot yu> for citing the full cause with its supporting + documentation and explanation. + +Thu Feb 22 15:59:12 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/config-win32-common.h: Change the comments around ACE_HAS_PENTIUM + to more accurately reflect its use and setting. With the switch to + QueryPerformanceCounter() for high-res timers on Windows, this + setting doesn't have any affect on timers, but it is used in + some CDR operations. This resolves Bugzilla #1410. + + * ace/OS_NS_time.inl (gethrtime): + * ace/gethrtime.cpp: Remove the comments directing the reader to a + description of using RDTSC on Windows. The comment is long gone, and + RDTSC isn't used on Windows/VC any longer anyway. + +Thu Feb 22 15:07:22 UTC 2007 Ciju John <johnc@ociweb.com> + + * bin/MakeProjectCreator/modules/GNUACEProjectCreator.pm: + * bin/MakeProjectCreator/templates/gnu.mpd: + Additions to support DDS specific build rules and versioning. + +Thu Feb 22 13:56:42 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Makefile.am: + Added Notification_Queue.{h,cpp,inl}: + + * ace/OS_NS_dirent.inl: + Added include when ACE_LACKS_ALPHASORT is defined + +Thu Feb 22 13:03:01 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * Merged changes from the bug_2815 branch. From revision 77182 to + revision 77315. + + Tue Feb 20 18:17:00 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * ace/ace_for_tao.mpc: + Add Notification_Queue.cpp to this file too. Thanks to Johnny + Willemsen for pointing this out. + + Mon Feb 19 21:53:58 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * ace/Intrusive_List.inl: + I learned a minute ago that std::swap() was kosher. Use it in + favor of ACE_Swap<>. + + Mon Feb 19 21:47:48 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * tests/run_test.lst: + * tests/Bug_2815_Regression_Test.cpp: + Adjusted the number of iterations so the test would pass on my + G4-based laptop without optimizations or inlining. The previous + numbers were just guesses anyway and I hope the new numbers will + result in no failures for the scoreboard. + + Mon Feb 19 21:29:32 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * ace/Notification_Queue.cpp: + Modified the purging algorithm. The algorithm used a temporary + list to store the elements not purged, so basically all nodes + were either moved to the free list or to the temporary list. + The new algorithm only removes the nodes that needs purging. + In the vast majority of the cases this is more efficient because + most nodes are not purged. + This resulted in a factor of 2 improvement for + tests/Bug_2815_Regression_Test, keep in mind that in this + program 1/16th of the nodes are purged. So in practice I would + expect even better results. + I also changed the order in which the free nodes are re-used, I + think LIFO order has a better chance of re-using the cache, but + I have no evidence or experiments to prove this. + + * ace/Intrusive_List.h: + * ace/Intrusive_List.cpp: + Renamed the remove_i() function to unsafe_remove() and promoted + its access to "public". The function is not safe to use in + general, thus the name, but it resulted in a factor of 2 + performance improvement for ACE_Notification_Queue. + + * tests/Bug_2815_Regression_Test.cpp: + Fixed memory leaks (in the test not the library) + + Mon Feb 19 20:32:34 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * ace/Notification_Queue.h: + * ace/Notification_Queue.inl: + * ace/Notification_Queue.cpp: + Re-factored to use ACE_Intrusive_List in the implementation. + This eliminates memory allocations during additions and removal + of elements. Also, many of the operations cannot fail, so the + code became smaller. + + * ace/Intrusive_List.h: + * ace/Intrusive_List.inl: + Add a swap() member function. + + Mon Feb 19 04:49:35 UTC 2007 coryan <coryan@atdesk.com> + + * ace/Notification_Queue.h: + * ace/Notification_Queue.cpp: + * ace/ace.mpc: + New class to encapsulate the implementation of a user-space + based notification queue. The code was duplicated in both + Select_Reactor.{h,cpp} and Dev_Poll_Reactor.{h,cpp} + + * tests/tests.mpc: + * tests/run_test.lst: + * tests/Notification_Queue_Unit_Test.cpp: + New unit test for the notification queue class. + + * ace/Dev_Poll_Reactor.h: + * ace/Dev_Poll_Reactor.cpp: + * ace/Select_Reactor_Base.h: + * ace/Select_Reactor_Base.cpp: + Refactored notification queue code to a new class + (ACE_Notification_Queue) + + * tests/Bug_2815_Regression_Test.cpp: + Add code to help with debugging. Basically my refactoring above + had at least one bug, and this test uncovered it. But it was + hard to debug because there was no single breakpoint to detect + when the failure condition was triggered. + + * tests: + Add new files to svn:ignore property. Also added some old + files. + + * tests/SSL: + * include/makeinclude: + Add missing files to svn:ignore + + Sun Feb 18 21:13:41 UTC 2007 coryan <coryan@atdesk.com> + + * tests/tests.mpc: + * tests/Bug_2815_Regression_Test.cpp: + Add new regression test for bug 2815. I have not added the test + to the automated test suite because (1) it fails, and (2) it is + a test to reproduce performance + +Thu Feb 22 12:58:42 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/compiler.bor: + Removed BCB4/5 support + +Thu Feb 22 11:19:59 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/OS_NS_dirent.h: + * ace/OS_NS_dirent.inl: + Added ACE_OS::alphasort(). alphasort() is a BSD extension found + on Linux as well. + + * ace/config-lynxos.h: + Added ACE_LACKS_ALPHASORT_PROTOTYPE. + + * ace/config-win32-common.h: + Added ACE_LACKS_ALPHASORT. + + * ace/os_include/os_dirent.h: + Added alphasort prototype. + + * tests/Dirent_Test.cpp: + Chaged to use use ACE_OS::alphasort(). + +Thu Feb 22 09:24:42 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Truncate.h (Truncate, truncate_cast): + Also renamed the Borland part + +Thu Feb 22 04:57:42 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * tests/NonBlocking_Conn_Test.cpp (test_connect): + + Fixed syntax error. s/sizeof ACE_TCHAR/sizeof (ACE_TCHAR)/. + +Wed Feb 21 21:46:18 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/config-pharlap.h: + * ace/config-WinCE.h: + * ace/OS_NS_stdio.inl: + * ace/OS_NS_unistd.inl: Changed ACE_LACKS_SETFILEPOINTEREX to + ACE_LACKS_WIN32_SETFILEPOINTEREX to be consistent with the other + ACE_*_WIN32_*things. + + * ace/config-pharlap.h: + * ace/config-WinCE.h: + * ace/NT_Service.{h cpp}: + * examples/NT_Service/nt_svc.h: + * examples/NT_Service/main.cpp: + Also added ACE_LACKS_WIN32_SERVICES for Pharlap and CE. This marks + a platform that is Windows, but doesn't have Windows services. + + * ace/Configuration.{h cpp}: + * ace/Registry.{h cpp}: Rather than exclude registry-using code using + ACE_HAS_PHARLAP, use a new setting ACE_LACKS_WIN32_REGISTRY. + + * ace/config-pharlap.h: Added ACE_LACKS_WIN32_REGISTRY. + + * protocols/ace/HTBP/HTBP_Environment.cpp: + * examples/Reactor/WFMO_Reactor/Registry_Changes.cpp: + * examples/Registry/test_registry_update.cpp: + * examples/Registry/test_registry_iterator.cpp: + * tests/Config_Test.cpp: Don't try registry stuff + if ACE_LACKS_WIN32_REGISTRY is set. + + * examples/Reactor/WFMO_Reactor/Window_Messages.cpp: Don't try this if + ACE_LACKS_MSG_WFMO is set. + + * apps/JAWS/server/IO.{h cpp}: + * apps/JAWS/server/HTTP_Handler.{h cpp}: + * apps/JAWS/server/HTTP_Server.{h cpp}: + * apps/JAWS2/JAWS/IO.{h cpp}: + * apps/JAWS2/JAWS/IO_Acceptor.{h cpp}: + * apps/JAWS2/JAWS/IO_Handler.{h cpp}: + * apps/JAWS3/jaws3/Asynch_IO.cpp: + * examples/APG/Proactor/HA_Proactive_Status.{h cpp}: + * examples/Connection/blocking/SPIPE-acceptor.{h cpp}: + * examples/Connection/blocking/test_spipe_acceptor.cpp: + * examples/Reactor/Proactor/post_completions.cpp: + * examples/Reactor/Proactor/simple_test_proactor.cpp: + * examples/Reactor/Proactor/test_cancel.cpp: + * examples/Reactor/Proactor/test_end_event_loop.cpp: + * examples/Reactor/Proactor/test_multiple_loops.cpp: + * examples/Reactor/Proactor/test_proactor.cpp: + * examples/Reactor/Proactor/test_proactor2.cpp: + * examples/Reactor/Proactor/test_proactor3.cpp: + * examples/Reactor/Proactor/test_timeout.cpp: + * examples/Reactor/Proactor/test_timeout_st.cpp: + * examples/Reactor/Proactor/test_udp_proactor.cpp: + * examples/Reactor/WFMO_Reactor/Talker.cpp: + Decide on whether or not to compile the asynch support classes + based on ACE_HAS_WIN32_OVERLAPPED_IO, not ACE_WIN32. ACE_WIN32 + is broader than that overlapped I/O-capable ones; Pharlap ETS + doesn't do overlapped I/O, for instance. + +Wed Feb 21 19:27:14 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/Array_Base.cpp (size): + + Simplified return value logic. + + * ace/Auto_Ptr.h (ACE_auto_ptr_reset): + + Added missing "brief" documentation. + + * ace/Caching_Strategies_T.inl: + * ace/Svc_Handler.cpp (init): + + Removed unnecessary uses of ACE_UNUSED_ARG macro. Just comment + out the parameter name instead. + + * ace/DLL_Manager.cpp (unload_all): + * ace/Functor.inl (operator()): + * ace/OS_NS_unistd.h (sbrk): + * ace/OS_NS_unistd.inl (sbrk): + * ace/Parse_Node.cpp: + * ace/Timer_Hash_T.cpp: + + Corrected type used to store pointer value. + s/ptrdiff_t/{u}intptr_t/. + + * ace/SOCK_Dgram.cpp: + + Fixed unused argument warning. + + * ace/String_Base_Const.cpp: + + Workaround npos constant run-time initialization problem in some + mixed shared/static linking conditions by making it easy for the + compiler to embed the value of the constant in the object file + at compile-time. + + * ace/Truncate.h (Truncate, truncate_cast): + + Deprecated ACE_Utils::Truncate() function name in favor of + ACE_Utils::truncate_cast(). The latter better reflects what the + function template actually does. + + * ace/Unbounded_Set.cpp: + * ace/Unbounded_Set.h: + * ace/Unbounded_Set.inl: + + Implement find() method in terms of begin/end iterators to prevent + search from looping indefinitely under some race condition + scenarios. + + Added missing const_iterator factory methods. + + Added missing const_iterator equality and inequality operators. + + * ace/OS_NS_unistd.inl: + * ace/Time_Value.inl: + * ace/Token_Request_Reply.inl: + * ace/FILE_IO.cpp: + * ace/Filecache.cpp: + * ace/High_Res_Timer.cpp: + * ace/Log_Record.cpp: + * ace/MMAP_Memory_Pool.cpp: + * ace/Mem_Map.cpp: + * ace/Message_Queue.cpp: + * ace/Message_Queue_T.cpp: + * ace/Name_Request_Reply.cpp: + * ace/Process.cpp: + * ace/SOCK_Dgram.cpp: + * ace/SOCK_IO.cpp: + * ace/Timer_Hash_T.cpp: + + s/ACE_Utils::Truncate/ACE_Utils::truncate_cast/g. The former is + deprecated. + + From Andrew Schnable + * ace/os_include/os_unistd.h (X_OK): + + Windows has no access() test for X_OK. Use R_OK instead. + + From Russell Mora + * ace/ACE.inl (hex2byte): + * ace/Capabilities.h (is_line, is_empty): + * ace/Capabilities.cpp (parse, fillent, is_entry, is_empty, is_line): + * ace/Get_Opt.cpp (long_option): + * ace/OS_NS_ctype.inl (ace_isalnum): + + Added ACE::ace_isalnum, ACE::ace_isdigit and ACE::ace_islower + and changed to use them where appropriate. This prevents the + narrow char versions being called with wide chars. + + * ace/Asynch_IO.cpp: + + Fixed preprocessor scoping issue in builds where + ACE_HAS_AIO_CALLS is undefined and versioned namspaces are + enabled. ACE_END_VERSIONED_NAMESPACE_DECL ended up being + preprocessed when it should not have. + + * ace/Configuration_Import_Export.cpp (export_config): + + Check the return value of fclose() is required because data may + have been buffered and will be fflush()d with the fclose(). + + * ace/Log_Msg.h: + * ace/Log_Msg.cpp: + + Changed message offset to correct type, ptrdiff_t, since it + contains a pointer difference value. + + * ace/Null_Semaphore.h: + + Added include to get definition of size_t. This is required for + static single-threaded builds on Darwin. + + * ace/OS_NS_stdio.inl (cuserid): + + Check for unexpected failure of getpwuid function; don't + dereference null pointer. + + * ace/Process.cpp (spawn): + + On Windows platforms a process handle is a pointer, not an int. + Corrected format specifiers accordingly. + + * ace/RB_Tree.cpp (remove_i): + + Fixes for Coverity REVERSE_INULL issues. Check for null pointer + before dereferencing. + + * ace/Recursive_Thread_Mutex.cpp (get_nesting_level): + + Reverted change that altered implementation for IA64 Windows + builds. The returned nesting level was incorrect. Use the + implementation used by both 32-bit and AMD 64 bit builds. + + * tests/Recursive_Mutex_Test.cpp (test_recursion_depth): + + Added test cases for the above problem. + + * ace/SOCK_Connector.cpp (shared_connect_finish): + + On Windows, in order to detect when the socket that has been + bound to is in TIME_WAIT we need to do the connect (which will + always return EWOULDBLOCK) and then do an + ACE::handle_timed_complete() (with timeout==0, i.e. poll). This + will do a select() on the handle which will immediately return + with the handle in an error state. The error code is then + retrieved with getsockopt(). + + Good sockets however will return from the select() with ETIME - + in this case return EWOULDBLOCK so the wait strategy can + complete the connection. + + * ace/Service_Manager.cpp (list_services): + + Type conversion - it is unlikely the name of the service is a + string bigger than 2GB (on 32-bit systems) so a static_cast<> is + safe here. + + * ace/Svc_Conf_Lexer.cpp (scan): + + Changed the loop indexing to allow loop variable to be the same + type as variable used to initialise the loop. This avoids + conversion warnings. + + (yylex): + + Fixed unused variable warning. + + * ace/UNIX_Addr.inl (set): + + Correct wide string to narrow string conversion. + + * ace/config-cxx-common.h: + + Eliminated copious amounts of warnings. + + * ace/config-linux.h (ACE_HAS_CLOCK_GETTIME): + * ace/config-linux-common.h (ACE_HAS_CLOCK_GETTIME): + + On linux clock_gettime/clock_settime are not available for + single-threaded builds. + + * ace/os_include/os_dlfcn.h (ACE_DEFAULT_SHLIB_MODE): + + Added DYNAMIC_PATH to the default flags used to call shl_load() + on HP 11.00 (PARISC1.1/32-bit) so that RPATH and SHLIB_PATH are + searched to find the requested .sl. + + * bin/MakeProjectCreator/templates/gnu.mpd (FORCED_IDL_STUBS): + + PCH source files are also dependent on the IDL compiler + generated files. + + * include/makeinclude/platform_hpux_aCC.GNU (SONAME): + + HP platform (non-g++) was missing the appropriate LDFLAGS for + building versioned shared libraries. Copied from the hpux + section of platform_g++_common.GNU. + + Disable warnings when 64 bit file offsets are enabled in HP + 11.11 builds. + + * include/makeinclude/platform_irix6.x_sgic++.GNU (AR, ARFLAGS): + + By default IRIX uses a prelinker to instantiate templates, thus + archives need to be created via CC (using the -ar option) so + that the required templates can be instantiated. This was the + cause of the missing symbols problems reported by users of the + static libraries. + + * include/makeinclude/platform_tru64_cxx.GNU: + + It seems that the default debug level in the Compaq C++ compiler + really goes to town when a template is instantiated - it seems + to add debug information for every scrap of code that the + template instantiation ever see resulting in debug template + instantiations that are up to 1000x bigger than their release + counterparts. Dialing the level back down to -g1 (-g2 is the + default) seems to avoid this issue and we still get debug + symbols for the template instantiation itself. + + * tests/NonBlocking_Conn_Test.cpp (test_connect): + + Fixed potential buffer overrun error. + + From Duane Binder + * ace/ace_wchar.h: + + Added ACE_TEXT_ macros needed by some apps. + + * ace/Process.inl (creation_flags): + * ace/Naming_Context.cpp: + * ace/Registry_Name_Space.cpp: + * ace/Registry_Name_Space.h: + + Use ACE_USES_WCHAR not UNICODE. Whether ACE uses Windows + wchar_t APIs is controlled by ACE_USES_WCHAR and not UNICODE. + + * include/makeinclude/platform_freebsd.GNU: + + Remove pthread and semaphore references for singled threaded + FreeBSD builds. + + * include/makeinclude/platform_linux.GNU (LIBS): + + Prevent single threaded linux builds from linking libpthread + (via librt). + + * include/makeinclude/rules.lib.GNU: + + Not all static libraries have template instantiations. Test if + $(TEMPINCDIR) exists before trying to add objects in an archive + library (was a problem on Tru64 with debug builds). + +Wed Feb 21 14:23:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/tao_orb_tests.lst: + Added bug 1592 regression + +Tue Feb 20 17:56:25 UTC 2007 Douglas C. Schmidt <schmidt@dre.vanderbilt.edu> + + * COPYING: Clarified how the COPYING file need to be included with + any code released. Thanks to Scott Mark <sjm at pobox dot com> + for helping clarify this. + +Wed Feb 21 08:35:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Swap.{h,cpp,inl}: + Removed, use std::swap instead of ACE_Swap + + * ace/ace.mpc: + * ace/ace_for_tao.mpc: + * ace/Makefile.am: + Removed Swap.{h,cpp,inl} + +Wed Feb 21 08:13:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * examples/IPC_SAP/SSL_SAP/SSL-server-poll.cpp: + * examples/IPC_SAP/SSL_SAP/SSL-server-simple.cpp: + * examples/IPC_SAP/SSL_SAP/SSL-server.cpp: + Fixed compile problems when wchar is enabled + +Tue Feb 20 19:15:30 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * tests/run_test.lst: + + Adding Object_Manager_Flipping_Test to the list of running + tests. + +Tue Feb 20 18:08:47 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * tests/Object_Manager_Flipping_Test.cpp: + * tests/tests.mpc: + + Adding new test verifying that after ACE::fini has been called, + the correct pointer value has been stored in TSS for the Service + Config. A subsequent call to ACE::init may SEGV + otherwise. Thanks to Bjoern Rasmussen <bjoern.d.rasmussen at + gmail dot com> for reporting it. + +Tue Feb 20 17:15:37 UTC 2007 Adam Mitz <mitza@ociweb.com> + + * ace/config-qnx-rtp.h: + * include/makeinclude/platform_qnx_rtp_gcc.GNU: + + Moved the version detection from the platform_macros file to the + config file. This allows us to use the QNX system-defined macro + _NTO_VERSION, which works in cross-compile environments. The old + method of reading the version number from the "uname -a" output is + not compatible with cross-compilation. + +Tue Feb 20 16:38:24 UTC 2007 William R. Otte <wotte@dre.vanderbilt.edu> + + * bin/MakeProjectCreator/templates/gnu.mpd + * bin/MakeProjectCreator/config/global.features + + Added logic in gnu template for new XSC custom type in CIAO, + and disabled xsc by default in global.features. + +Tue Feb 20 16:12:57 UTC 2007 Steve Huston <shuston@riverace.com> + + * tests/Bug_2820_Regression_Test.cpp: ACEify use of auto_ptr. + +Tue Feb 20 16:06:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/os_include/os_stdlib.h: + Simplified the ACE_RANDR_TYPE define setting + + * ace/OS_NS_stdio.inl: + * ace/SOCK_Dgram.cpp: + * ace/WFMO_Reactor.cpp: + Simplified Borland C++ checks, no need to check for the version + number + +Tue Feb 20 15:54:20 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * ace/OS_NS_Thread.cpp: + + Reverted my change to check PTHREAD_STACK_MIN. I just realized + that it may not be defined as an integral value (i.e., Solaris). + +Tue Feb 20 15:46:06 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * ace/OS_NS_Thread.cpp: + + Added ACE_UNUSED_ARG(id) for RTEMS and modified code to check + that size (which is of type size_t) is less than PTHREAD_STACK_MIN + only if PTHREAD_STACK_MIN is greater than zero. + + * ace/OS_NS_stdlib.cpp: + + Fixed a compile error on VxWorks where an attempt to cast + ACE_thread_t to ACE_RANDR_TYPE failed (since ACE_thread_t is char* + on VxWorks). + + * ace/SOCK_Dgram.cpp: + + The iov_len element of the iovec struct is unsigned on RTEMS (as + it is with Linux and Borland C++), so use the same #if check to + avoid code that causes warnings. + +Tue Feb 20 15:06:04 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * ace/SOCK_Dgram.cpp: + + The iov_len element of the iovec struct is unsigned on Linux (as + it is with Borland C++), so use the same #if check to avoid code + that causes warnings. + + * ace/Containers_T.cpp: + * ace/Timer_Heap_T.cpp: + * ace/UPIPE_Stream.cpp: + * ace/UUID.cpp: + + Added static_cast<> to avoid conversion warnings. + +Tue Feb 20 15:00:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * m4/compiler.m4: + Fixed support for --enable-symbol-visibility. This fixes bugzilla + bug 2771, thanks to Waba <waba-ace at waba dot be> for reporting + this. + +Tue Feb 20 14:43:00 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * ASNMP/asnmp/pdu.cpp: + * apps/JAWS2/JAWS/Cache_Manager.cpp: + + Added a check for self assignment. + + * ace/INET_Addr.cpp: + * tests/Multicast_Test_IPV6.cpp: + * examples/Web_Crawler/Command_Processor.cpp: + + Removed redundant code to avoid "statement is unreachable" + warnings. + + * examples/APG/Naming/Temperature_Monitor.cpp: + * examples/APG/Naming/Temperature_Monitor2.cpp: + * examples/APG/Streams/Answerer.cpp: + * examples/C++NPv1/Logging_Server.cpp: + * examples/C++NPv1/Process_Per_Connection_Logging_Server.cpp: + * examples/C++NPv1/Thread_Per_Connection_Logging_Server.h: + + Added ACE_NOTREACHED around unreachable code to avoid "statement + is unreachable" warnings. + + * examples/Connection/blocking/SPIPE-acceptor.cpp: + + Reverted a bug that was introduced many years ago where threads + would be spawned and then the main thread would exit. + + * performance-tests/SCTP/SOCK_SEQPACK_clt.cpp: + * performance-tests/SCTP/SOCK_SEQPACK_srv.cpp: + * performance-tests/SCTP/SOCK_STREAM_srv.cpp: + + Modified to only compile code if ACE_HAS_SCTP is defined. + + * tests/INET_Addr_Test.cpp: + + Check addr.get_host_addr() for non-zero before passing it to + ACE_OS::strcmp(). + +Tue Feb 20 03:08:26 UTC 2007 Carlos O'Ryan <coryan@atdesk.com> + + * tests/tests.mpc: + * tests/Bug_2820_Regression_Test.cpp: + A new test to demonstrate the problems in bug 2820. + +Mon Feb 19 23:13:07 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/WFMO_Reactor.h: Clarify description of ok_to_wait(). + + * ace/WFMO_Reactor.cpp (ok_to_wait): Since WinCE can't use the + waitAll arg to atomically wait for both the lock and ok_to_wait + event, use a 2-step process to get both individually. This avoids + entering wfmo either without the ok_to_wait event or without + the lock. Thanks to Craig Johnson for reporting this. Also see: + Mon Oct 16 14:24:25 UTC 2006 Steve Huston <shuston@riverace.com> + +Mon Feb 19 21:02:55 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/Atomic_Op.inl: + * ace/OS_NS_unistd.inl: Adjust code for less-than-modern Win32 systems + such as Pharlap ETS to work with 64 bit file offsets and the lack + of interlocked exchange-add. + +Mon Feb 19 18:33:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/SSL/SSL_Context.h: + Layout change + + * ace/SSL/SSL_Context.inl (check_context): + Set the verify mode correctly, fixes bugzilla bug 2819 + +Mon Feb 19 12:00:12 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/tao_other_tests.lst: + Added Secure_Invocation test, this tests basic functionality but + seems to fail at this moment because of some broken support in + TAO + +Fri Feb 16 23:48:55 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/Service_Config.cpp: + + ACE_INLINE makes sense in .inl files (thanks Steve) + +Fri Feb 16 13:38:04 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/Dynamic_Service_Base.cpp: + * ace/Dynamic_Service_Dependency.cpp: + * ace/Service_Config.h: + * ace/Service_Config.inl: + * ace/Service_Config.cpp: + * ace/Service_Gestalt.h: + * ace/Service_Gestalt.inl: + * ace/Service_Gestalt.cpp: + + Reverting my changes since "Tue Feb 13 15:35:54 UTC 2007" as it + has been causing problems, in particular with borland and + no-threads builds. + +Fri Feb 16 12:47:03 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/Service_Config.cpp (global): Removed the ACE_INLINE qualifier + from global() - if it's in the .cpp file, it shouldn't be marked + as inline, and isn't marked thusly in the .h file. Fixes unsats. + + * ace/Service_Config.h: Fixed spelling error. + +Thu Feb 15 20:04:01 UTC 2007 Ossama Othman <ossama_othman at symantec dot com> + + * ace/OS_NS_stdlib.cpp (mkstemp_emulation): + + Include process and thread ID in random number generator seed + value. Addresses problems where the same seed value would end + up being used for different processes and/or threads, resulting + in identical filenames being generated. + + Addressed potential race condition on platforms that do not + inline the value of ACE_Numeric_Limits<>::max(). + +Thu Feb 15 17:53:53 UTC 2007 Krishnakumar B <kitty@nospam.invalid.domain> + + * THANKS: Check-in rules 101: Check-in files that you have + modified. Don't ever forget this basic rule. + +Thu Feb 15 13:50:58 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/CDR_Stream.h: + * ace/Connector.h: + * ace/DEV_IO.h: + * ace/FIFO_Recv.h: + * ace/FIFO_Send.h: + * ace/FIFO_Send_Msg.h: + * ace/FILE_IO.h: + * ace/INET_Addr.h: + * ace/Local_Memory_Pool.h: + * ace/Malloc_Base.h: + * ace/MEM_IO.h: + * ace/Mem_Map.h: + * ace/MEM_SAP.h: + * ace/MEM_Stream.h: + * ace/Message_Block.h: + * ace/MMAP_Memory_Pool.h: + * ace/OS_NS_unistd.h: + * ace/Pipe.h: + * ace/Process_Manager.h: + * ace/QoS/SOCK_Dgram_Mcast_QoS.h: + * ace/Service_Types.h: + * ace/Shared_Memory_Pool.h: + * ace/SOCK_Dgram.h: + * ace/SOCK_Dgram_Mcast.h: + * ace/SOCK_IO.h: + * ace/SOCK_SEQPACK_Association.h: + * ace/SPIPE_Stream.h: + * ace/Stream.h: + * ace/Task.h: + * ace/Thread.h: + * ace/Thread_Manager.h: + * ace/UPIPE_Stream.h: + Doxygen improvements + +Thu Feb 15 13:38:58 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/ATM_Connector.h: + * ace/DEV_Connector.h: + * ace/DEV_IO.h: + * ace/Dev_Poll_Reactor.h: + * ace/FILE_Connector.h: + * ace/Hash_Map_Manager_T.h: + * ace/IOStream_T.h: + * ace/LOCK_SOCK_Acceptor.h: + * ace/Map_Manager.h: + * ace/MEM_IO.h: + * ace/MEM_SAP.h: + * ace/MEM_Stream.h: + * ace/Message_Block.h: + * ace/Message_Queue.h: + * ace/Message_Queue_T.h: + * ace/Module.h: + * ace/Proactor.h: + * ace/Reactor.h: + * ace/Reactor_Impl.h: + * ace/Select_Reactor_T.h: + * ace/Sig_Handler.h: + * ace/Signal.h: + * ace/SOCK.h: + * ace/SOCK_Acceptor.h: + * ace/SOCK_Dgram.h: + * ace/SOCK_IO.h: + * ace/SOCK_SEQPACK_Acceptor.h: + * ace/SOCK_SEQPACK_Association.h: + * ace/SPIPE_Connector.h: + * ace/Stream.h: + * ace/Svc_Handler.h: + * ace/Task_Ex_T.h: + * ace/Task_T.h: + * ace/Thread_Manager.h: + * ace/TLI_Acceptor.h: + * ace/TLI_Connector.h: + * ace/Token.h: + * ace/WFMO_Reactor.h: + * ace/XTI_ATM_Mcast.h: + Doxygen improvements + +Thu Feb 15 13:12:01 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * examples/OS/Process/imore.cpp: + + Removed an enclosing block around code that caused issue with + SunCC 5.8. There was a comment claiming that the block was needed + so that the ACE_Process_Options could elegantly close the file + handlers that it kept, however the ACE_Process_Options object does + not close the file handlers. The ACE_Process does the handle + closing. + +Thu Feb 15 05:41:11 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/Service_Gestalt.inl: + * ace/Service_Gestalt.cpp: + + Moved find off-line because it requires a full declaration of + ACE_Service_Repository, which is unavailable when the inl file + is compiled (inlining on) + +Wed Feb 14 23:05:16 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/Service_Config.h: + * ace/Service_Config.inl: + * ace/Service_Config.cpp: + * ace/Service_Gestalt.h: + * ace/Service_Gestalt.cpp: + + Moved the ACE_Unmanaged_Singleton usage to the implementation + file and eliminated the unnecesary #includes, aiming to reduce + the footprint. + +Wed Feb 14 19:56:14 UTC 2007 Adam Mitz <mitza@ociweb.com> + + * ace/Service_Config.h: + s/TAO_Export/ACE_Export/ + +Wed Feb 14 19:21:58 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/tao_orb_tests.lst: + Added bug 2809 regression + +Wed Feb 14 17:33:22 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/Service_Config.h: + Fixing a compile problem with g++ and borland compiler, which + treat members of private nested classes as "hidden" even when + the enclosing type has been exported. + +Wed Feb 14 12:59:58 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Acceptor.h: + Doxygen improvement + +Wed Feb 14 11:31:58 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/os_include/os_stdio.h: + * ace/os_include/os_unistd.h: + Move ACE_STDIN, ACE_STDOUT and ACE_STDERR from stdio to unistd + file and use STDIN_FILENO, STDOUT_FILENO and STDERROR_FILENO if + these are defined yet, this way we use the same defines and header + files as in the posix specification. This fixes bugzilla bug 2731. + Thanks to Abdul Sowayan <abdullah dot sowayan at lmco dot com> + for reporting this. + + * ace/Truncate.h: + Added another const specialization for Borland C++ + +Wed Feb 14 11:03:58 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * tests/Process_Manager_Test.cpp: + Added test code from bugzilla 2797, thanks to Yves Alloyer + <alloyer at ipanematech dot com> for reporting this. + +Wed Feb 14 10:06:58 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/tao_other_tests.lst: + Added TAO/orbsvcs/tests/Bug_2800_Regression/run_test.pl + +Wed Feb 14 08:53:58 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Active_Map_Manager_T.h: + * ace/Cache_Map_Manager_T.h: + * ace/CDR_Size.h: + * ace/CDR_Stream.h: + * ace/Configuration.h: + * ace/Dev_Poll_Reactor.h: + * ace/Future.h: + * ace/Future_Set.h: + * ace/Hash_Cache_Map_Manager_T.h: + * ace/High_Res_Timer.h: + * ace/INET_Addr.h: + * ace/IO_SAP.h: + * ace/IPC_SAP.h: + * ace/Map_T.h: + * ace/MEM_SAP.h: + * ace/Message_Block.h: + * ace/Msg_WFMO_Reactor.h: + * ace/Multihomed_INET_Addr.h: + * ace/Priority_Reactor.h: + * ace/Reactor.h: + * ace/Reactor_Impl.h: + * ace/Recursive_Thread_Mutex.h: + * ace/Registry.h: + * ace/Semaphore.h: + * ace/Shared_Memory_MM.h: + * ace/Shared_Memory_SV.h: + * ace/SOCK_SEQPACK_Association.h: + * ace/Thread_Mutex.h: + * ace/Timeprobe_T.h: + * ace/WFMO_Reactor.h: + Doxygen improvements + + * ace/Truncate.h: + Fixed compile problems with Borland C++, thanks to Ossama for + giving a solution. + +Tue Feb 13 20:33:58 UTC 2007 Krishnakumar B <kitty@dre.vanderbilt.edu> + + * THANKS: Added gzeleniy@gmail.com to the THANKS list. + +Tue Feb 13 15:35:54 UTC 2007 Iliyan Jeliazkov <iliyan@ociweb.com> + + * ace/Service_Config.{h,inl,cpp}: + + Refactoring to simplify the TSS usage, reduce overhead and + eliminate a memory leak. + + * ace/Service_Gestalt.cpp: + + Fixed a memory leak in Service_Type_Dynamic_Guard dtor. Updated + logging. + + * ace/Dynamic_Service_Base.cpp: + * ace/Dynamic_Service_Dependency.cpp: + + Updated to use SC::instance () to better express the contract + for these classes. They simply don't need to be aware whether + they are using a global, or a TSS-local gestalt . + +Tue Feb 13 14:44:06 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/config-vxworks5.x.h: + Removed VxWorks 5.4.2 support, we haven't build for it in years + +Tue Feb 13 12:58:13 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * ACEXML/common/Exception.h: + * ACEXML/common/Exception.cpp: + * ACEXML/common/SAXExceptions.h: + * ACEXML/common/SAXExceptions.inl: + * ACEXML/common/SAXExceptions.cpp: + + Changed the print() and message() methods to const. There was no + reason for thease to be non-const and it makes it possible to + catch const references of the ACEXML exception types. + +Mon Feb 12 20:24:04 UTC 2007 Steve Huston <shuston@riverace.com> + + * examples/Logger/Acceptor-server/server_loggerd.cpp (handle_input): + * examples/Logger/simple-server/Logging_Handler.cpp (handle_input): + Log the host name of the peer, not "<localhost>", as previous code + did; also make sure to put it in ACE_TCHAR form. See also + Thu Feb 8 23:09:59 UTC 2007 Douglas C. Schmidt <schmidt@dre.vanderbilt.edu> + +Mon Feb 12 19:20:06 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/CORBA_macros.h: + Fixed several macros which where using () around the exception + to be thrown + +Mon Feb 12 18:10:42 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/Name_Request_Reply.cpp (timeout): + * netsvcs/lib/Time_Request_Reply.cpp (timeout, decode): + Truncate 64-bit time value down to the platform's time_t size. + +Mon Feb 12 15:18:42 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/fuzz.pl: + + Check for the use of ACE_THROW_SPEC in TAO or CIAO code only. + It's usage within ACE is still ok in some situations. + +Mon Feb 12 14:44:06 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/fuzz.pl: + Removed old todo's, extended msvc ver check + + * ace/Codeset_IBM1047.cpp: + * ace/Functor.cpp: + * ace/Functor.inl: + * ace/Hash_Map_Manager_T.cpp: + * ace/Hash_Multi_Map_Manager_T.cpp: + * ace/Sched_Params.cpp: + * ace/Sched_Params.inl: + Converted to doxygen style documentation + + * bin/doxygen-convert-h.pl: + Only handle source files + +Mon Feb 12 14:38:06 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/CORBA_macros.h: + Use ACE_LACKS_DEPRECATED_MACROS to make it possible to build without + all the emulated exception macros. Will make sure people can't use them + anymore in core code without noticing that + + * ace/Global_Macros.h: + Just define ACE_THROW_SPEC again as about a week ago, we need a fuzz check + to make sure people don't use this anymore in TAO + + * ace/Reactor.h: + Layout change + + * tests/Thread_Manager_Test.cpp: + Layout change to debug message + +Mon Feb 12 14:35:25 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/doxygen-convert-h.pl: + + Allow filenames or directories. If a directory is passed, it is + recursively searched for any file. + +Mon Feb 12 13:57:06 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * etc/*.doxygen: + Removed ACE_THROW_SPEC + +Sat Feb 10 23:13:06 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/Svc_Conf_Lexer.cpp (ace_yy_buffer_state): Disable VC8 warning + C4351 (input_ elements will be default-initialized) for this + block. The new behavior is good and desireable, so the warning + noise can be silenced in this case. + + * ace/Timer_Hash_T.cpp (reschedule, schedule_i): When calculating the + hash value, truncate the seconds value down to size_t range because + the modulo value used to hash is a size_t. Avoids a compile warning + about losing data when time_t is larger than size_t. + +Fri Feb 9 21:40:02 UTC 2007 Krishnakumar B <kitty@nospam.invalid.domain> + + * ace/High_Res_Timer.inl: + * ace/High_Res_Timer.cpp: + + Removed ACE_WIN32 specific code in ACE_High_Res_Timer to reflect + the change in the default value of ACE_HR_SCALE_CONVERSION. + Removed uses of ACE_HRTIME_CONVERSION and use + ACE_Utils::Truncate instead. + + * ace/OS_NS_time.h: + + Changed the check for defining ACE_HRTIME_CONVERSION and + ACE_HRTIME_TO_U64 from ACE_HRTIME_T_IS_BASIC_TYPE to + ACE_LACKS_UNSIGNEDLONGLONG_T. + + * tests/High_Res_Timer_Test.cpp: + + Updated the test to use ACE_Utils::Truncate instead of + static_cast which caused 0 to be output when time_t is 64-bit. + +Fri Feb 9 12:19:57 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/rm_exception_macros.pl: + + Fixed a bug where the end of the ACE_THROW_SPEC would not be + recognized. + +Thu Feb 8 23:09:59 UTC 2007 Douglas C. Schmidt <schmidt@dre.vanderbilt.edu> + + * examples/Logger/Acceptor-server/server_loggerd.cpp (handle_input): + * examples/Logger/simple-server/Logging_Handler.cpp (handle_input): + * examples/Logger/client/logging_app.cpp (ACE_TMAIN): Fixed these + programs to use the proper (new) formatting for ACE_Log_Record. + Thanks to Mouna Seri <seri at crhc dot uiuc dot edu> for + reporting this. + +Thu Feb 8 19:40:30 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/Makefile.am: + * ace/ace.mpc: + * ace/Time_Request_Reply.{h cpp}: + * netsvcs/lib/Makefile.am: + * netsvcs/lib/lib.mpc: + * netsvcs/lib/Time_Request_Reply.{h cpp}: Moved Time_Request_Reply + from ace to netsvcs; it's only used from netsvcs and is only useful + as an example of some ACE usage. Real time synch needs should be + met using NTP. + + * ace/Name_Request_Reply.{h cpp}: Changed the sec_timeout_ transfer + member from ACE_UINT32 to ACE_UINT64 to accomodate 64-bit time_t. + + * netsvcs/lib/TS_Clerk_Handler.h: + * netsvcs/lib/TS_Server_Handler.h: Time_Request_Reply.h now is included + from netsvcs/lib, not ace. + + * netsvcs/lib/Time_Request_Reply.{h cpp}: Changed the time-passing + API arguments in the constructor and time() methods from ACE_UINT32 + to time_t. The on-wire protocol buffer fields that carry seconds + values between clerk and server were changed from ACE_UINT32 to + ACE_UINT64 to accomodate 64-bit time_t values. + + * NEWS: Explained the user-visible ramifications of these changes. + +Thu Feb 8 09:23:50 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/MMAP_Memory_Pool.cpp: + Removed empty else statement. + + * include/makeinclude/platform_linux.GNU: + Fixed TCL/TK settings + +Thu Feb 8 08:03:57 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/config-openbsd.h: + Removed ACE_LACKS_CLEARERR and added ACE_HAS_3_PARAM_READDIR_R. + +Wed Feb 7 23:04:49 UTC 2007 Wallace Zhang <zhang_w@ociweb.com> + + * bin/tao_other_tests.lst: + + Added two existing Notify Service examples into the autotest list. + +Wed Feb 7 21:01:39 UTC 2007 Steve Huston <shuston@riverace.com> + + * ACE-INSTALL.html: Added a bit of info about building for National + Instruments' LabVIEW RT which uses Pharlap ETS. In particular, if + building for LabVIEW's Pharlap, add #define ACE_PHARLAP_LABVIEW_RT + to your config.h file along with #include "ace/config-pharlap.h" + to adjust for slight differences between stock Pharlap and + LabVIEW RT's variant. + + * ace/config-pharlap.h: Pharlap doesn't supply InterlockedExchangeAdd + and doesn't do SetFilePointerEx (needs ACE_LACKS_SETFILEPOINTEREX). + + * ace/OS_NS_Thread.cpp (sched_params): Don't attempt SetPriorityClass + on Pharlap - it's not defined. + + * ace/OS_NS_Thread.inl (thr_getprio): For Pharlap ETS, use + EtsGetTimeSlice() to determine the scheduling policy; if the time + slice is anything other than 0, it's multitasking ACE_SCHED_OTHER. + + * ace/OS_NS_stdio.inl (cuserid): Pharlap doesn't go GetUserName() so + make it ACE_NOTSUP_RETURN. + (adjust_flock_params): Use SetFilePointer() instead of + SetFilePointerEx(), even with 64 bit offsets, if + ACE_LACKS_SETFILEPOINTEREX is set. + + * ace/OS_NS_unistd.inl (pwrite): For Windows platforms without + overlapped I/O, don't rely on availability of SetFilePointerEx - + platforms that have it also have overlapped I/O. In this case, use + SetFilePointer with appropriate call differences for 32/64 bit + file offsets. + + * ace/Pagefile_Memory_Pool.{h inl cpp}: Pharlap doesn't do this class. + + * ace/Configuration.{h cpp}: + * ace/Registry.{h cpp}: The Windows registry is not available in + Pharlap. + +Wed Feb 7 14:44:23 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * bin/rm_exception_macros.pl: + + Added the ability to remove the ACE_THROW_SPEC's completely. + +Tue Feb 6 16:44:00 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Global_Macros.h: + For the moment expand ACE_THROW_SPEC to nothing, we want to deprecate + this in TAO and want to see the impact of this on the tests + +Tue Feb 6 15:36:00 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/Time_Value.inl: Remove the #if around #include "ace/Truncate.h" + because ACE_Utils::Truncate is needed in a case that's not in a + ACE_HAS_TIME_T_LONG_MISMATCH block. + +Mon Feb 5 19:58:39 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/Log_Record.cpp (operator>>): Use ACE_Utils::Truncate to truncate + the received 64-bit time seconds to time_t. + +Mon Feb 5 17:16:20 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/config-win32-msvc-8.h: Added a new setting for Windows VC8, + ACE_HAS_TIME_T_LONG_MISMATCH, to capture the case that requires + a replacement timeval def. See the following for details: + Thu Feb 1 23:46:27 UTC 2007 Steve Huston <shuston@riverace.com> + This change isolates just the desktop VC8 compiler for the change, + avoiding compilers that don't need it and/or can't deal with + the code. + + * ace/Time_Value.{h inl}: Change code that uses a replacement timeval + from ACE_WIN32 to ACE_HAS_TIME_T_LONG_MISMATCH. + +Fri Feb 2 14:45:00 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Dynamic_Service_Base.cpp: + Fixed compile error + +Fri Feb 2 14:32:06 UTC 2007 Adam Mitz <mitza@ociweb.com> + + * bin/auto_run_tests.pl: + Modified default behavior (if no -a, -o, -t, -C options are given) + so that the script looks for TAO in $TAO_ROOT as well as + $ACE_ROOT/TAO and similarly for CIAO. This change supports the peer + style layout. + +Fri Feb 2 14:08:00 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/rm_exception_macros.pl: + Also zap ACE_THROW_RETURN from application code, that is also + deprecated now + +Fri Feb 2 12:03:00 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/TSS_T.cpp: + Fixed merge error + +Fri Feb 2 11:43:00 UTC 2007 Simon Massey <sma@prismtech.com> + + * include/makeinclude/platform_hpux_aCC.GNU: + Supress 3.xx warnings about non-return from functions + (Anoyingly it doesn't seem to think that a throw + terminates the execution flow of a function.) + +Fri Feb 2 10:47:55 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/Proactor.cpp: + Layout change. + + * ace/config-lynxos.h: + Enabled POSIX_SIG_PROACTOR and made some cleanups. + + * ace/os_include/net/os_if.h: + Removed LynxOS specific include. + + * tests/Proactor_Test.cpp: + * tests/Proactor_Test_IPV6.cpp: + Enabled these tests again even if ACE_HAS_BROKEN_SIGEVENT_STRUCT + is defined. + +Fri Feb 2 10:31:14 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/CORBA_macros.h: + Removed workaround for ACE_THROW_RETURN for HP aCC compiler. This is + only needed for the HP 3.x compiler and we can suppress just this + warning. When this has been build on the scoreboard we will + zap the usage of this macro. + + * ace/Dynamic_Service.cpp: + * ace/Dynamic_Service_Base.cpp: + * ace/Service_Config.{h,cpp}: + * ace/Service_Gestalt.cpp: + * ace/TSS_T.cpp: + Layout changes + + * ace/Thread_Manager.cpp: + Const improvement + +Fri Feb 2 06:56:14 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/ACETAOCIAO.style: + Configuration file for profactor (see www.profactor.co.uk), a + code beautifier tool that we want to use to reformat the code + in ACE/TAO/CIAO. + +Fri Feb 2 06:48:40 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/POSIX_CB_Proactor.h: + * ace/config-aix-5.x.h: + * ace/config-freebsd.h: + * ace/config-integritySCA.h: + * ace/config-macosx-panther.h: + * ace/config-macosx-tiger.h: + * ace/config-macosx.h: + * ace/config-rtems.h: + * ace/config-sunos5.4-g++.h: + * ace/config-sunos5.4-sunc++-4.x.h: + * ace/config-sunos5.5.h: + * ace/config-vxworks5.x.h: + * ace/config-vxworks6.2.h: + * ace/config-vxworks6.3.h: + * ace/config-vxworks6.4.h: + + Removed unnecessary sigval_t typedef and feature test macro + ACE_LACKS_SIGVAL_T + +Thu Feb 1 23:46:27 UTC 2007 Steve Huston <shuston@riverace.com> + + * NEWS: Note change to ACE_Log_Record constructor. + + * ace/Log_Record.h (constructor): The "long time_stamp" argument was + changed from long to time_t to correspond with what's expected. + + * ace/Log_Record.{h inl cpp}: The secs_ value was changed from + ACE_UINT32 to time_t and is handled as a ACE_CDR::LongLong in the + CDR insertion and extraction operators. The encode() and decode() + methods are deprecated and will be removed in a future version of + ACE. + + * ace/Message_Block.cpp (priority_status): Cast sec() value to + suseconds_t size to avoid truncation warnings. + + * ace/Time_Value.{h inl cpp}: Use a private struct to manage time + internally for Windows since Windows' struct timeval uses long + values internally instead of the standard time_t/suseconds_t + that the class API uses (and POSIX uses in timeval). When a + real OS-defined timeval is needed, build it on the fly for Windows. + + * tests/High_Res_Timer_Test.cpp: When logging ACE_Time_Value-obtained + sec/usec values, truncate them down to u_int size. Avoids passing + 64-bit values to ACE_Log_Msg when only 32 bits will be taken from + the arg vector for a %u format spec. Fixes crashing of this test + on Win/VC8. + + * examples/APG/Naming/Temperature_Monitor.cpp: + * examples/APG/Naming/Temperature_Monitor2.cpp: Treat time as an int + instead of a time_t; makes the logic easier even though it'll break + sometime in 2038. + + * examples/ASX/UPIPE_Event_Server/event_server.cpp: Unify use of time_t + to hold time values from ACE_OS::time(). + + * examples/IPC_SAP/SOCK_SAP/CPP-inclient.cpp: When measuring messages + per second, truncate seconds down to a 'long' to avoid truncation + warnings at compile time from converting to double. + + * examples/Reactor/Misc/notification.cpp (svc): Initialize an + ACE_Time_Value with secs / 2 instead of getting the seconds, + multiplying into usecs, then dividing by 2. Avoids data truncation + issues. + + * examples/Reactor/Multicast/Log_Wrapper.cpp (log_message): Cast + result of time() to long to insert in the multicast buffer. Avoids + truncation warning. Will stop recording a legit result sometime in + 2038 but the receiver isn't looking at the time field. + + * apps/gperf/src/Gen_Perf.cpp: + * ASNMP/tests/test_config.h (randomize): + * examples/Reactor/WFMO_Reactor/Multithreading.cpp: + * protocols/ace/RMCast/Link.cpp: + * protocols/ace/RMCast/Simulator.cpp: + * protocols/examples/TMCast/Member/member.cpp: + * tests/Auto_Event_Test.cpp: + * tests/Handle_Set_Test.cpp: + * tests/Recursive_Mutex_Test.cpp: + * tests/Semaphore_Test.cpp: + * tests/Thread_Mutex_Test.cpp: + * tests/Timer_Queue_Test.cpp: Cast time_t to u_int for use as an + ACE_OS::srand() seed - we just want some bits, not to fit it back in + range with ACE_Utils::Truncate. + +Thu Feb 1 11:21:32 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/POSIX_CB_Proactor.h: + * ace/POSIX_CB_Proactor.cpp: + * ace/Proactor.cpp: + * ace/config-freebsd.h: + * ace/config-lynxos.h: + + Introduced a new feature test macro ACE_HAS_BROKEN_SIGEVENT_STRUCT + to make POSIX_CB_Proactor compile on FreeBSD 7.0-CURRENT. + + * tests/Proactor_Test.cpp: + * tests/Proactor_Test_IPV6.cpp: + + Modified due to changes above. + +Thu Feb 1 08:05:14 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Message_Queue.{h,cpp,inl}: + Fixed compile errors in the NT Message Queue when build without + threading under windows + + * ace/OS_NS_Thread.cpp: + Layout changes + +Wed Jan 31 23:06:42 UTC 2007 Adam Mitz <mitza@ociweb.com> + + * ace/config-qnx-rtp-62x.h: + + Replaced ACE_LACKS_CUSERID with ACE_HAS_ALT_CUSERID to get the ACE + library to provide us with a cuserid() function. + + * ASNMP/asnmp/asn1.cpp: + * apps/drwho/global.h: + * examples/IPC_SAP/SOCK_SAP/C-inclient.cpp: + * examples/IPC_SAP/SOCK_SAP/C-inserver.cpp: + * examples/IPC_SAP/SOCK_SAP/CPP-inclient.cpp: + * examples/IPC_SAP/SOCK_SAP/CPP-inserver.cpp: + * examples/Logger/Acceptor-server/server_loggerd.cpp: + * examples/Logger/client/logging_app.cpp: + * examples/Logger/simple-server/Logging_Handler.cpp: + * examples/Reactor/Multicast/Log_Wrapper.cpp: + * examples/Reactor/Multicast/server.cpp: + * netsvcs/lib/Name_Handler.cpp: + + Replaced htonl, htons, ntohl, and ntohs with the appropriate + ACE_XTOYZ macro (ACE_HTONL, for example). QNX doesn't have htonl and + friends without adding extra #includes. + +Wed Jan 31 19:03:37 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * ace/Basic_Types.h: + + Added constants ACE_FLT_MIN and ACE_DBL_MIN to provide consistency + with ACE_FLT_MAX AND ACE_DBL_MAX. + + * ace/CDR_Base.h: + * ace/CDR_Base.cpp: + + Enhanced the ACE_CDR::LongDouble class which is used on platforms + where the native long double is not 16 bytes. + + Provided operators to support multiplication, division, addition + and subtraction as well as conversion to the native long double. + + An assign() method is also provided to set the value of the + ACE_CDR::LongDouble. Since this class will be stored in a union + (if used as an IDL type), it can not have a non-trivial constructor + or assignment operator. The existing + ACE_CDR_LONG_DOUBLE_INITIALIZER macro is used to initialize the + LongDouble and a new macro, ACE_CDR_LONG_DOUBLE_ASSIGNMENT, is + used to assign to the LongDouble in a portable fashion. + + * bin/tao_orb_tests.lst: + + Added the new TAO LongDouble test. + +Wed Jan 31 17:30:46 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/config-hpux-11.00.h: + * include/makeinclude/platform_hpux_aCC.GNU: If stdcpplib=0 is + selected at build time, explicitly set -AP compile option to use + old C++ libraries and no std namespace. Also set + ACE_LACKS_NUMERIC_LIMITS when -AP is set - there is a numeric_limits + but not in std. Note that ACE_Array_Map won't build with stdcpplib=0 + but pre-ACE-5.5 legacy code will still build. + + Also removed the HP-UX 10.x remnants and pre-3.0 aC++ support + from platform_hpux_aCC.GNU. Could probably safely remove more old + compiler support (anything older than 3.56 probably) but I played + it a little conservative for now. + +Wed Jan 31 17:22:11 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * protocols/examples/examples.mwc: + + Removed the implicit usage. It is not necessary since each + example has a .mpc file. + + * protocols/examples/example_base.mpb: + + Removed this file. It was not needed. + +Wed Jan 31 16:55:23 UTC 2007 Adam Mitz <mitza@ociweb.com> + + * ace/config-qnx-rtp-62x.h: + + Removed #define ACE_HAS_INLINED_OSCALLS as it seems to be both rarely + used (only oddball platforms like unixware and other QNX versions + have it) and also incompatible with inline=0 builds. + Also added new config macros ACE_LACKS_STRPTIME (June 2006) and + ACE_HAS_3_PARAM_READDIR_R (December 2006). + These modifications have been tested with QNX 6.3 but from the docs + on the QNX web site I believe they apply to QNX 6.2 as well. + +Wed Jan 31 15:44:54 UTC 2007 Olli Savia <ops@iki.fi> + + * ace/POSIX_CB_Proactor.h: + * ace/POSIX_CB_Proactor.cpp: + Use sigval (which is SUS defined) instead of sigval_t. If this + change doesn't break on any platform I'll remove unnecessary + typedef and related stuff. + +Wed Jan 31 13:31:14 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * bin/fuzz.pl: + Removed obsolete checks for environment macros + + * ace/Recycable.h: + Removed empty brief doxygen tag + + * ace/OS_NS_unistd.inl: + * ace/config-rtems.h: + Renamed ACE_HAS_BROKEN_SZ_PAGESIZE to ACE_HAS_NOTSUP_SC_PAGESIZE + +Wed Jan 31 13:21:14 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/platform_hpux_gcc.GNU: + Added improvement for GCC on HPUX IA64 + +Wed Jan 31 10:17:14 UTC 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/platform_hpux_gcc.GNU: + Improved support for GCC on HPUX IA64 + +Wed Jan 31 07:50:44 UTC 2007 Krishnakumar B <kitty@nospam.invalid.domain> + + * ace/OS_NS_Thread.cpp (ACE_OS::sched_params): Check if the + ACE_id_t id that is being passed in is equal to ACE_SELF. If + so, pass ::GetCurrentProcessId() to ::OpenProcess() instead of + id. ACE_SELF is defined to be 0 on ACE_WIN32; process id 0 is + the zero-page process, and we certainly don't want to attempt to + set it to REALTIME_PRIORITY_CLASS :-) + +Wed Jan 31 03:25:51 UTC 2007 Krishnakumar B <kitty@nospam.invalid.domain> + + * ace/High_Res_Timer.h: + + Removed the special case default definition of + ACE_HR_SCALE_CONVERSION to ACE_ONE_SECOND_IN_MSECS on Windows; + added a #ifndef around the default definition in case the user + wishes to change it. This was done to prevent useless values + like 0 showing up in calculations using ACE_Basic_Stats when the + time intervals are of the order of a few microseconds. Most of + the current machines have a QueryPerformanceCounter() + implementation that returns a value of atleast 3 MHz. So it + should be completely safe to use the definition of + ACE_ONE_SECOND_IN_USECS as the default value for + ACE_HR_SCALE_CONVERSION on contemporary x86 machines running + Windows. + + * ace/Message_Block.inl: + * ace/Message_Block.cpp: + + Moved the definition of + ACE_Dynamic_Message_Strategy::priority_status() from + Message_Block.inl to Message_Block.cpp. It was a big function + and it currently has code that causes a warning with VC8 on + Windows. + +Tue Jan 30 19:21:50 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * ace/Module.h: + + Removed the trailing comma from the last element of the enum. + +Tue Jan 30 16:26:43 UTC 2007 William R. Otte <wotte@dre.vanderbilt.edu> + + * bin/sets-manager.py: + + Modified to pin the version of MPC at the time of branching/tagging + under modules/ACE/MPC. Note: This does not branch OR tag MPC, + it simply causes the branch/tag to check out a specific + revision of MPC. + +Tue Jan 30 13:51:32 UTC 2007 Chad Elliott <elliott_c@ociweb.com> + + * ace/Module.h: + * ace/Module.cpp: + + There was no way to distinguish between flags not being set and + flags to indicate that tasks should not be deleted. I have + changed the M_DELETE_NONE enum value to 4 and added a new enum + value for use in flag initialization. With this change, the + close() method can determine if it needs to use the flags passed + in or the internal flags. This fixes Bugzilla #2497. + +Tue Jan 30 11:35:14 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Countdown_Time.{h,cpp}: + Readded destructor, missed that it updates the max_wait_time + passed to the constructor + +Tue Jan 30 10:44:14 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * include/makeinclude/wrapper_macros.GNU: + When pipes is set to 1, also add -pipe to the CCFLAGS + + * include/makeinclude/platform_rtems.x_g++.GNU: + Set pipes to 1, set DCCFLAGS + +Tue Jan 30 10:14:14 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Unbounded_Set.cpp: + Initialise pointer with 0. + + * ace/OS_NS_unistd.inl (getpagesize): + * ace/config-rtems.h: + RTEMS does define _SC_PAGESIZE, but using sysconf with _SC_PAGESIZE + results in -1. Added a new define ACE_HAS_BROKEN_SC_PAGESIZE to be + able to disable the usage of sysconf (_SC_PAGESIZE) for rtems 4.6, + the ::getpagesize() function returns a correct value. + +Tue Jan 30 09:46:14 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * tests/OS_Test.cpp: + Added a test for ACE_OS::getpagesize(), it should return a value + bigger then 0 else the memory pools don't work + +Tue Jan 30 00:25:00 UTC 2007 Steve Huston <shuston@riverace.com> + + * ace/config-hpux-11.00.h: If the user sets RWSTD_NO_NAMESPACE to + keep a lot of the standard C++ library out of the std namespace, + avoid setting ACE_USES_STD_NAMESPACE_FOR_STDCPP_LIB, and also + hoist everything else from std into the global namespace. This is + necessary because RW wasn't consistent with using their internal + _RWSTD_NO_NAMESPACE and the user-settable RWSTD_NO_NAMESPACE and + some things end up in std and some not; so get the ones that ended + up in std out to global with the rest of them. + +Mon Jan 29 19:48:14 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Based_Pointer_Repository.cpp: + * ace/Basic_Stats.cpp: + * ace/Codeset_Registry.cpp: + * ace/FILE.cpp: + * ace/FILE_Addr.cpp: + Layout improvements + +Mon Jan 29 19:07:14 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * etc/ciao.doxygen: + * etc/ciao_config_handlers.doxygen: + * etc/ciao_DAnCE.doxygen: + Removed CIAO_NAMESPACE + +Mon Jan 29 18:58:14 2007 Johnny Willemsen <jwillemsen@remedy.nl> + + * ace/Argv_Type_Converter.cpp: + * ace/Asynch_Pseudo_Task.cpp: + * ace/Cached_Connect_Strategy_T.cpp: + Layout changes, result of experimenting with a code + beautifier + + * ace/config-rtems.h: + Improved rtems port, added support for building with + threads=0 + + * ace/config-sco-5.0.0.h: + Removed commented out define + + * ace/config-vxworks6.4.h: + putenv fails with VxWorks 6.4 with a pentium target + + * ace/Countdown_Time.{h,cpp}: + Use bool, removed return value from most methods, it was + meaningless, removed destructor, not needed to update members + in a destructor + + * ace/Dev_Poll_Reactor.h: + Export Token_Guard, fixes bugzilla 2786, thanks to + <sailzeng at tencent dot com> for reporting this + + * OS_NS_pwd.inl (getpwnam_r): + Initialize pointer with 0, simplified this a little + + * docs/ACE-guidelines.html: + Updated example with doxygen style of documentation + Sun Jan 28 19:59:14 2007 William R. Otte <wotte@dre.vanderbilt.edu> * ACE version 5.5.6 released. |