diff options
Diffstat (limited to 'ACE/ChangeLogs/ChangeLog-98a')
-rw-r--r-- | ACE/ChangeLogs/ChangeLog-98a | 6588 |
1 files changed, 0 insertions, 6588 deletions
diff --git a/ACE/ChangeLogs/ChangeLog-98a b/ACE/ChangeLogs/ChangeLog-98a deleted file mode 100644 index c5daa4d7ee1..00000000000 --- a/ACE/ChangeLogs/ChangeLog-98a +++ /dev/null @@ -1,6588 +0,0 @@ -Wed Apr 29 18:55:13 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE version 4.5 released. - -Wed Apr 29 12:25:05 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-lynxos.h: added ACE_POLL_IS_BROKEN so that - ACE_OS::sleep () uses select () instead of poll (). - - * tests/Map_Manager_Test.cpp: added a ssize_t loop index that's - used to avoid signed/unsigned comparisons. - -Tue Apr 28 23:08:10 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Hash_Map_Manager.cpp: Bugs fixed. - - * tests/Map_Manager_Test.cpp: Added assertion tests on the number - of iterations to ensure the test does run correctly. - - * tests/Map_Manager_Test.cpp (main): Disabled the VERBOSE_LITE - Log_Msg flag before starting the real test and reenabled it - after the tests. - -Tue Apr 28 06:49:50 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE-INSTALL.html,os-patches/linux-patches.html: updated - Linux glibc-2 info. ACE works with glib-2.0.7-7, unmodified. - - * ace/config-g++-common.h: added - ACE_LACKS_STATIC_DATA_MEMBER_TEMPLATES with egcs. egcs 1.0.2 - won't compile netsvcs/lib without it. - - * include/makeinclude/platform_vxworks5.x_ghs.GNU: fixed location - of parenthesis in ppc check. Thanks to Brian Mendel - <bmendel@mdc.com> for reporting this. - - * ace/High_Res_Timer.cpp: initialize global_scale_factor_ to 1000 - on Linux. - -Mon Apr 27 07:18:54 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Service_Types.cpp: Changes to use typedef - ACE_Service_Object_Exterminator to define <gobbler>. Thanks to - Chuck Gehr for reminding this. - - * ace/Service_Object.h (ACE_Service_Object_Ptr): - * netsvcs/servers/main.cpp (main): Added more comments on how to - use ACE_Service_Object_Ptr with ACE_SVC_INVOKE correctly. - - * ace/OS.h (ACE_Service_Object_Exterminator): Define the type as - extern "C" function pointer to make MVS happy. Thanks to Chuck - Gehr for pointing this out. - - * ace/Service_Object.i (ACE_Service_Object_Ptr): Delete the - underlying Service_Object after calling its fini() method. - Thanks to Tom Arbuckle <arbuckle@uran.informatik.uni-bonn.de> - for pointing this out. - -Sun Apr 26 21:50:07 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Svc_Conf_l.cpp,etc/Svc_Conf_[ly].cpp.diff,etc/README: - updated the patch files. Thanks to J. Russell Noseworthy - <rnosewor@objectsciences.com> for reporting problems with - them. - -Sun Apr 26 01:05:20 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.43 released. - -Sat Apr 25 17:47:31 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.h (ACE_FACTORY_DECLARE): Added explicit casting for - gobbler to prevent some weird compiler from complaining too - much. - - * netsvcs/lib/Server_Logging_Handler_T.cpp - (ACE_Server_Logging_Acceptor_T): We can't initialize the - <scheduler_> with a temporary object. Its content gets deleted - right after the constructor exits. - -Sat Apr 25 11:55:01 1998 Alexander Babu Arulanthu <alex@merengue.cs.wustl.edu> - - * ace/Containers.h (class ACE_Double_Linked_List): Added - comments.(<class T>, the parameter for this class should have - T* next_ and T* prev_ fields in it. - - * ace/Containers.h (class ACE_Double_Linked_List): Corrected the - documentation for the insert_head and insert_tail functions. - -Fri Apr 24 22:36:37 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Service_Config.cpp (process_directive): added - ACE_UNUSED_ARG (directive). - -Fri Apr 24 18:40:08 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Asynch_Acceptor.cpp (handle_accept): Fixed this method so - that it has better flexibility and error checking. Also, added - hooks so that users can do validation of remote connections. - Also, a flag is checked before accept is reissued. Thanks to - Jason Katz <Jason@rtswireless.com> for reporting these upgrades. - -Fri Apr 24 16:33:14 1998 Steve Huston <shuston@riverace.com> - - * ace/OS.h: ACE_FACTORY_DEFINE - in the generated gobble function, - use the indicated object type to do the delete so 1) the aC++ - compiler won't complain, and 2) so the service object is - destructed properly. - - * tests/Env_Value_Test.cpp: Made it compile on Solaris. - -Fri Apr 24 15:45:38 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-sunos5.5-*.h: set ACE_THR_PRI_FIFO_DEF to 30, because - sched_get_priority_min is not implemented on Solaris 2.5.x. - -Fri Apr 24 09:12:39 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * tests/Env_Value_Test.cpp: Made it UNICODE safe. - - * tests/version_tests/version_tests.dsw: - * tests/version_tests/OrdMultiSet_Test.dsp: Added new project - file. - -Fri Apr 24 09:09:49 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-linux-common.h: added ACE_HAS_4_4BSD_SENDMSG_RECVMSG. - Thanks to Sandro Doro <doros@aureus.sublink.org> for recommending - this. - -Fri Apr 24 04:12:39 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * tests/version_tests/version_tests.dsw: - * tests/version_tests/OrdMultiSet_Test.dsp: Added new project - file. - -Thu Apr 23 22:52:20 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * tests/tests.dsw: Somehow, Thread_Mutex_Test.dsp got removed from - the workspace. - - * tests/Thread_Mutex_Test.dsp: Update include/lib path. - -Fri Apr 24 00:12:32 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.42 released. - -Thu Apr 23 22:28:02 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Service_Config: Began adding support for the new service - configurator feature that will make it possible to dynamically - configure services that are read in from the command-line. - Thanks to Karlheinz for requesting this. - - * ace/Parse_Node.cpp (open_handle), - ace/Svc_Conf.l: The line for the service object for the service - configurator contains the DLL specifications. Previously, it - was possible to specify either the absolute path name of the DLL - or only the name of the DLL itself. For the latter, the service - configurator is searching for the DLL using the PATH environment - variable. If we used the absolute path name it was not - possible to use environment variables in the string. It is now - possible to do this for Win32 using their %foo% notation. - Thanks to Andreas Schuelke <Andreas.Schuelke@med.siemens.de> for - providing these fixes. - - * etc/Svc_Conf_y.cpp.diff, - etc/Svc_Conf_l.cpp.diff, - ace/Makefile: Change the YY* symbols to ACE_YY* symbols to - avoid clashes if we link with other flex/yacc generated code. - - * apps/Gateway/Gateway/gatewayd.cpp, - apps/Gateway/Peer/peerd.cpp: Added a check to see if the - svc.conf file exists before trying to parse the arguments. - Thanks to Dani Flexer <danif@ivory-sw.com> for reporting this. - - * apps/Gateway/Peer/Options.cpp (parse_args): Added the "C:" - qualifier to the get_opt string. Thanks to Dani Flexer - <danif@ivory-sw.com> for reporting this. - - * ace/Activation_Queue.h: Extended ACE_Activation_Queue so that it - can be a proper base class by making the destructor virtual and - putting the methods into the protected section. Thanks to Garry - Brother <gmbroth@romulus.ncsc.mil> for this. - - * ace/XtReactor.cpp: Commented out most of the logic if we're - running on Win32 because (1) it doesn't work correctly since - HANDLEs are not ints and (2) there's no X windows on Win32 - anyhow... - - * ace/XtReactor.h: Made ACE_XtReactor be ACE_Export'd. Thanks to - Kirill Rybaltchenko <Kirill.Rybaltchenko@cern.ch> for reporting - this. - -Thu Apr 23 13:54:33 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Malloc.i (free): added ACE_UNUSED_ARG (ptr) if ACE_NDEBUG. - - * include/makeinclude/platform_chorus.GNU: include dtool/htgt-cf.rf - from $(MERGEDIR) instead of $(CLASSIXDIR). Thanks to - Wei Chiang <chiang@horizon.nokia.com> for this update. - -Wed Apr 22 22:13:57 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/ace_bcc.ide: Added a new IDE for the Borland C++ compiler. - Thanks to Valik Solorzano Barboza <valik@geodan.nl> for - this. - - * ace/Msg_WFMO_Reactor.cpp (wait_for_multiple_events): Fixed a - missing semi-colon. Thanks to Jorn Jensen <jornj@funcom.com> - for reporting this. - -Wed Apr 22 21:53:12 1998 David L. Levine <levine@cs.wustl.edu> - - * config-linux-kcc.h: added #define ACE_SIZEOF_LONG_DOUBLE - to 12 on i386. Thanks to Ben Eng <ben@jetpen.com> for - reporting this. - -Tue Apr 21 18:38:22 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ace_ce_dll.dsp: Removing a deprecated /windowsce:noconvert - linker flag. - -Mon Apr 20 15:39:42 1998 Steve Huston <shuston@riverace.com> - - * ace/Handle_Set.cpp: Access fds_bits elements using an fd_mask - pointer rather than array indexing of the fds_bits array itself. - HP-UX 11 plays some games with the fd_mask type - fd_mask is defined - as an int32_t, but the fds_bits is an array of longs. This makes - plainly indexing through the array by hand tricky, since the FD_* - macros treat the array as int32_t. So the bits are in the right - place for int32_t, even though the array is long. This, they say, - is to preserve the same in-memory layout for 32-bit and 64-bit - processes. So, we play the same game as the FD_* macros to get - the bits right. See /usr/include/sys/_fd_macros.h. - On all other systems, this amounts to practically a NOP, since this - is what would have been done anyway, without all this type jazz. - -Sat Apr 18 13:59:55 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Naming_Context.h: removed trailing ';' at end of - ACE_FACTORY_DECLARE. - - * ace/OS.*,Sched_Params.{h,cpp}: added support on Solaris - for setting the LWP priority. It is necessary to set the - LWP priority for bound threads in the Real-Time scheduing - class. Thread priority is essentially ignored. See the - pthread_setschedparam man page for more info. Thanks to - Fred Kuhns <fredk@arl.wustl.edu> for helping track this problem - down. - - There is one drawback: to set the LWP priority, in ACE, - a thread must set its own priority. There are examples - in performance-tests/Misc/preempt.cpp: - - High_Priority_Task::svc (void) - { - // On Solaris 2.5.x, the LWP priority needs to be set. - // This is the ACE way to do that . . . - ACE_OS::thr_setprio (priority_); - - I don't know of a way to set the LWP priority for another - thread. The problem is that there is no way that I know of - to find the LWP ID of another (bound) thread. A thread can - find the ID of its own LWP using _lwp_self (). - - * ace/config-sunos5*,README: added ACE_HAS_PRIOCNTL and - ACE_NEEDS_LWP_PRIO_SET. - - * ace/config-osf1-4.0.h: added ACE_HAS_PRIOCNTL and - ACE_LACKS_PRI_T. - - * performance-tests/Misc/preempt.cpp (svc): set LWP priority - (on Solaris) by having the thread set its own priority. - This test now passes on Solaris 2.5.1! - - * ace/High_Res_Timer.cpp (reset): initialize by assignment with - 0 instead of using memset. Thanks to Carlos for noticing this. - - * ace/Object_Manager.cpp (ACE_Object_Manager ctor): allocate - registered_objects_ after the allocation of ACE_STATIC_OBJECT_LOCK, - just to be safe. - -Sat Apr 18 01:28:36 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Timeprobe.h (class ACE_Timeprobe): Changed ACE_Timeprobe to - use ACE_OS::gethrtime(). - -Sat Apr 18 01:14:30 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/OS.cpp (start and stop): start and stop should only be done - when ACE_Countdown_Time::max_wait_time_ is non-zero. - - * ace/Acceptor.cpp (open): Only set the reactor if calls to open() - and register_handler() succeeds. - -Fri Apr 17 21:00:21 1998 Steve Huston <shuston@riverace.com> - - * ace/config-hpux11.h: Changes and comments around the - ACE_DEFAULT_BASE_ADDR definition for 64-bit HP-UX 11. - - * ace/Naming_Context.cpp - * ace/Malloc_T.cpp - * ace/Memory_Pool.cpp: Hacks to work around bugs in HP-UX aC++ in - 64-bit mode on HP-UX 11. - - * ace/OS.i: Addition of a facility that allows config.h to specify - additional flags for ACE_OS::mmap, above what's set by the caller. - It's done using a new definition, ACE_OS_EXTRA_MMAP_FLAGS. - This was added primarily to allow setting of the MAP_ADDR32 flag - on HP-UX 11 if the ACE_DEFAULT_BASE_ADDR is set in the 32-bit range - to allow 32-bit and 64-bit programs to map the same region. - -Fri Apr 17 11:40:38 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.cpp (invoke): Should check against null thr_desc, not - thr_desc->flag (). Thanks to "Erickson, Jack (CICG - NY GFX)" - <JErickson@exchange.ml.com> for reporting this. - - * ace/OS.h (ACE_Service_Object_Exterminator): Typedef of ACE - factory cleanup funciton. - (ACE_FACTORY_{DECLARE,DEFINE}): Added an extra argument to pass - back destructor for the object created by the factory. Changed - the ACE_FACTORY_DEFINE to generate this destructor function to - pair with factory function so that we can deallocate the object - in the same heap it was created. - (ACE_SVC_INVOKE): This will now pass in 0, which means we don't - want to use the destructor function. - - * ace/Parse_Node.{h,cpp} (symbol): Added an extra argument to get - back the destructor function from ACE_FACTORY. Only - Function_Node actually uses it. - (ACE_Function_Node::symbol): Initialize the Service_Object with - added destructor function ptr so it will get destructed in the - "right place". - - * ace/Service_Config.{h,cpp} (load_static_svcs): Invoke the - factory method with added argument to get back the destructor - function and then, pass it in to ace_create_service_type - function. - (ACE_Service_Manager): Changed to use the more generic - ACE_FACTORY_DEFINE than using ACE_SVC_FACTORY_DEFINE. - (ACE_SERVICE_ALLOCATOR): Must match with signature of the - new factory method. - - * ace/Service_Type.{h,cpp} - (ACE_Service_Object_Type,ACE_Service_Type_Impl): Keep the - destruction function in these object, if use provides one, then - use it to destruct the object, otherwise, just use delete. - - * ace/Naming_Context.{h,cpp}: Changed to use - ACE_FACTORY_{DEFINE,DECLARE}. - - * ace/Object_Manager.cpp: Changed the signature of - Service_Manager. - - * ace/Svc_Conf.h: Changed the signature of - ace_create_service_type. - - * ace/Svc_Conf_y.cpp: - * ace/Svc_Conf.y: Try to get the destruction function from - factories in svc_location. - (ace_create_service_type): Pass in the destruction function if - provided one. - -Fri Apr 17 08:31:38 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Signal.cpp (dispatch): Fixed a typo. Thanks for David - Levine's help pointing this out. - -Thu Apr 16 21:12:48 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Signal.cpp: There was a major portability violation in the - ACE_Sig_Handler::dispatch() method, which was previously - acquiring a mutex lock. This is NOT permitted in the POSIX - signal handler spec. In addition, the implementation had other - nasty consequences related to dynamic allocation of memory in - the signal handler, which has a bad habit of breaking - non-reentrant uses of "new". The consequence of all this is - that you really shouldn't be modifying signal handlers via the - Reactor in multiple threads of control. In general, threads and - signals are just plain evil, so they are best addressed via - sigwait() anyhow... - - Thanks to Sumedh and Naga for first tracking this down. - -Thu Apr 16 20:51:42 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-linux-pthread.h: added ACE_HAS_THREADS. - ace/config-irix6.x-g++.h: added ACE_HAS_THREADS and ACE_MT_SAFE. - Thanks to Dave Tallman <tallman@acsys.com> for reporting these. - -Thu Apr 16 04:14:53 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Timeprobe: - - This class is used to instrument code. This is accomplished by - inserting time probes at different location in the code. - ACE_Timeprobe then measures the time difference between two time - probes. - - This class provides a lightweight implementation for measuring - the time required to execute code between two time probes. When - a time probe executes, it records the time, the id of the - calling thread, and an event description. The event description - can either be an unsigned long or a string (char *). If string - are used, care must be taken cause only pointer copies are done - and the string data is *not* copied. - - The recorded time probes can then be printed by calling - print_times(). If you have used unsigned longs as event - descriptions in any of your time probes, you must provide an - event description table that maps the unsigned longs to readable - strings. This map is a simple array of strings, and the event - number is used as the index into the array when looking for the - event description. If you have only used strings for the event - description, this map is not necessary. - -Wed Apr 15 23:01:08 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Thread_Manager.cpp (exit): Don't delete the ACE_Thread_Exit - object of the thread here. ACE_Thread_Manager::exit will be - called back again when performing TSS cleanup. Thanks to Dave - Tallman <tallman@acsys.com> for the bug report. - -Wed Apr 15 17:12:14 1998 Steve Huston <shuston@riverace.com> - - * ace/config-hpux-11.x-hpc++.h: Explicitly set ACE_SIZEOF_LONG based - on compiler switch for processor/memory model. - -Wed Apr 15 16:25:38 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/WFMO_Reactor.cpp (handle_signal): Changed break to return 0. - - * ace/OS.cpp (uname): We have to make sure that the size of - (processor + subtype) is not greater than the size of - name->machine. So we give half the space to the processor and - half the space to subtype. The -1 is necessary for because of - the space between processor and subtype in the machine name. - - Thanks to Andrew Marchewka - (Andrew_Marchewka@SRT.CCMAIL.compuserve.com) for suggesting - these changes. - -Wed Apr 15 08:52:52 1998 David L. Levine <levine@cs.wustl.edu> - - * tests/Reactor_Exceptions_Test (handle_input): removed - ACE_NOTREACHED wrapper around return statement at end. - Now that the previous statement is a function call, the - compiler (cxx) doesn't detect that the statement can't - be reached. - -Wed Apr 15 01:03:42 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS.i (ACE_SOCKCALL_RETURN): Added a work-around for a - Borland 5.02 compiler bug. The key turns out to be the - throw-away intermediate assignment statement to correctly keep - the value of WSAGetLastError(). Thanks to Dave Tallman - <tallman@acsys.com> for reporting this. - -Tue Apr 14 23:35:19 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Profile_Timer.cpp (elapsed_rusage): The differences of user - time and system time for GETRUSAGE were not calculated. - - * performance-tests/Synch-Benchmarks/Options.cpp (print_results): - Added dump code for ACE_HAS_GETRUSAGE. - - * ace/ace_{dll,lib}.dsp: Removed Service_Main.cpp from the - projects. We don't really need them in ACE DLL. - - * ace/Service_Main.cpp: Removed the file from CVS. Lots of - Service_Config examples can be found all round ACE directories - so we really don't need it here anymore. - -Tue Apr 14 20:30:53 1998 Steve Huston <shuston@riverace.com> - - * ace/OS.i (mutex_init, cond_init): Added sections to properly - handle error conditions on platforms which claim - ACE_HAS_PTHREADS_1003_DOT_1C. - -Tue Apr 14 16:39:42 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ACE.cpp (ldfind): Fixed a miscalculated string size. - Thanks to Jason Katz <Jason@rtswireless.com> for reporting the - bug. - -Tue Apr 14 14:33:33 1998 David L. Levine <levine@cs.wustl.edu> - - * tests/Thread_Manager_Test.cpp (main): don't try to suspend/resume - DCE threads, because they don't support it. Added a start barrier, - to ensure that worker threads set up their signal handlers before - the main thread tries to signal them. Drop DEFAULT_ITERATIONS from - 100000 to 10000 because 100000 takes a long time on a 200 MHz - Pentium. - -Tue Apr 14 12:59:04 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * tests/Hash_Map_Manager_Test.cpp: Changed the consts to - size_t, as well... - -Tue Apr 14 11:40:23 1998 David L. Levine <levine@cs.wustl.edu> - - * tests/Hash_Map_Manager.cpp: made type, size_t, of String_Table_size - explicit, instead of using the default type of int. - -Tue Apr 14 10:10:46 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.41 released. - -Tue Apr 14 07:19:25 1998 David L. Levine <levine@cs.wustl.edu> - - * bin/make_release (create_kit): ACE+TAO instead of ACE_TAO. - Thanks to Doug for reporting this. - - * ace/OS.h (ACE_Thread_Adapter): added friend class declaration - to avoid g++ warning: only defines a private destructor and - has no friends. - - * ace/OS.cpp (ACE_Thread_Adapter::invoke): only declare local - thr_desc if ACE_HAS_WIN32 and ACE_HAS_MFC. - -Tue Apr 14 00:46:05 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.40 released. - -Mon Apr 13 19:19:13 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS: Made the destructor of ACE_Thread_Adapter private to - keep people like Russ from abusing this class... ;-) - - * ace/OS.cpp (invoke): We were incorrectly deleting this and then - using one of its fields! Thanks to the ever-vigilant Russ - Noseworthy <rnosewor@objectsciences.com> for reporting this. - - * ace/Log_Record.cpp (print): Added VERBOSE_LITE support to the - code so we don't spew out quite so much information. - - * ace/Log_Msg.cpp (log): Replaced all calls to - ACE_Log_Record::print() so that they pass the - ACE_Log_Msg::flags_ parameter. - - * ace/Log_Record.h: Changed the 'int verbose' parameter to 'u_long - verbose_flag' parameter so that we can pass in the - ACE_Log_Msg::flags_ and check to see what they are enabled with! - - * tests/test_config.h: Changed VERBOSE to VERBOSE_LITE so we don't - have quite so much junk printed out for the tests! - - * ace/Log_{Priority,Record}: Added a new flag called VERBOSE_LITE - that isn't quite as verbose as VERBOSE (it only prints things - that can change between calls to the ACE_Log_Msg::log() method). - In addition, also changed the ACE_Log_Record::print() method so - it prints the string format for the ACE_Log_Priority types. - This is a lot more legible than printing the numbers. - - * ace: Removed all uses of ACE_HAS_WILDCARD_BIND. This is now the - default behavior. If there's some weird platform that DOESN'T - have this behavior, we'll deal with that by setting the - ACE_LACKS_WILDCARD_BIND flag. - - * ace/OS.cpp (invoke): Removed the (void *) cast on delete this. - It wasn't necessary. Thanks to Steve Huston and David Levine - for reporting this. - -Mon Apr 13 19:12:49 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/DEV_Connector.i (ACE_DEV_Connector): - * ace/Asynch_IO.cpp (accept,addresses): UNICODE changes. - -Mon Apr 13 18:05:26 1998 Steve Huston <shuston@riverace.com> - - * tests/Thread_Manager_Test.cpp: Disabled the suspend/resume action - for Pthreads platforms. - -Mon Apr 13 10:18:59 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/High_Res_Timer.i (hrtime_to_tv): in tv.usec calculation, - moved the cast to long to after the division by global_scale_factor_. - This is necessary for compatibility with ACE_U_LongLong. - - * bin/make_release (get_versions): use $KIT instead of hard-coded - ACE, so that it works with TAO. (tag): translate dots to - underscores, because CVS doesn't allow dots in tags. Thanks to - Doug and Hans Rohnert <rohnert@scr.siemens.com> for reporting these. - -Sun Apr 12 21:11:53 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.h: Prevent using ACE_HAS_MOSTLY_UNICODE_APIS without UNICODE. - - * ace/OS.cpp (thr_create): - * ace/Thread_Manager.cpp (spawn_i): Removed directive - ACE_NO_THREAD_ADAPTER and its related code. - -Sat Apr 11 22:20:51 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * tests/Thread_Manager_Test.cpp (main): Don't check execute result - of kill_grp statement if we are running on Win32 platforms. - thr_kill is not supported there. - -Sat Apr 11 20:00:12 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.39, released Sat Apr 11 20:00:12 1998. - -Sat Apr 11 19:58:56 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * Nanbor removed a bunch of *.MAK files, which were quite large - and taking up lots of space. - -Sat Apr 11 18:58:24 1998 Steve Huston <shuston@riverace.com> - - * ace/INET_Addr.cpp: Fixed missing paren. - - * netsvcs/servers/servers4.mdp: Added ..\lib to include directories - for Release build. - -Sat Apr 11 13:28:19 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.38, released Sat Apr 11 13:28:19 1998. - -Sat Apr 11 12:46:08 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/config-mvs.h: Added some ACE_SIZEOF_* macros for MVS. - Thanks to Chuck Gehr <gehr@sweng.stortek.com> for reporting - this. - - * ace/INET_Addr.cpp: Added a newline and a better explanation of - what went wrong if an error occurs. Thanks to Luca for - reporting this. - -Fri Apr 10 02:42:35 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Log_Msg.{h,cpp} (log_hexdump): - * ace/ACE.{h,cpp} (format_hexdump): The buffer to be hex-dumped should - always be of type (const char *) no matter we have UNICODE - system strings or not. - - * ace/OS.{h,cpp} (ACE_CE_Bridge): This new class bridges - a text-based ACE programs with a default window that prints out - the default ACE_Log_Msg output. - - CE always uses non-static object manager and we replace the - original main function with a well know entry that the main - window program can call back from to invoke the ACE program in a - separate thread. - - * ace/Log_Record.{h,cpp}: Added support for passing strings as - windows messages. - - * ace/Log_Msg.cpp (log): Specialized CE support. - - * windoezCE: The template windows program that receives ACE - messages. See CE-Status or README file for more info. - -Fri Apr 10 00:41:23 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * tests/Reactor_Exception_Test: Tidied the test up a bit. Thanks - to Bob Laferriere <laferrie@gsao.med.ge.com> for suggesting - this. - - * tests/test_config.h: Enabled ACE_Log_Msg::VERBOSE since it gives - more information that can be useful for debugging. - -Fri Apr 10 00:04:03 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Proactor.cpp (post_completion): Added a new method to the - Proactor. It can be used to post a result to the completion port - of the Proactor. If errors occur, the result will be deleted by - this method. If successful, the result will be deleted by the - Proactor when the result is removed from the completion port. - Therefore, the result should have been dynamically allocated and - should be orphaned by the user once this method is called. - - Changed the timer mechanism to use this new method rather than - posting to the completion port directly. - - Thanks to Jason Katz <Jason@rtswireless.com> for the - improvements. - -Thu Apr 9 17:31:47 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * Updated all the Makefile dependencies and also added the - appropriate CVS strings to all Makefiles. - - * examples/IPC_SAP/SOCK_SAP: Improved both the CPP-unserver.cpp - and CPP-inserver.cpp programs so they work identically wrt - concurrency and logic. - - * ace/LSOCK_Stream.cpp: Fixed the get_remote_addr() and - get_local_addr() methods to work correctly. Apparently, - getpeername() doesn't work appropriately for UNIX domain - sockets, so we simply have both methods call getsockname(). - Thanks to Chuck Gehr for pointing out this problem. - - * ace/FILE_IO.cpp (get_local_addr): The RHS has an additional - field that is the FILE name, which did not get copied. Added an - ACE_dynamic_cast() to do the right thing here. Thanks to Ganesh - Pai for reporting this. - - * ace/Malloc*: Factored out all the code in ACE_Static_Allocator - into a new non-templatized base class called - ACE_Static_Base_Allocator so that we don't trigger multiply - defined symbols when we use this in TAO. Thanks to Carlos for - this very insightful trick! - - * tests/Thread_Manager_Test.cpp (main): Added ACE_ASSERT - statements so that we bailout if things go awry. Thanks to - Robert J. Laferriere <laferrie@gsao.med.ge.com> for reporting - this. - -Thu Apr 09 18:06:53 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/High_Res_Timer.i (hrtime_to_tv): Cast the parameter for usec - to long. It can be fixed because of Darrell's vigilance. - -Thu Apr 09 17:55:21 1998 Steve Huston <shuston@riverace.com> - - * ace/config-hpux-10.x.h, config-hpux11.h, config-aix-4.2.x.h: - Added ACE_HAS_WILDCARD_BIND. - -Thu Apr 09 16:43:26 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/ACE.*: Changed *_version() functions to be non-inlined. - After this change, inclusion of ace/Version.h can be moved into - ACE.cpp. This change helps as all of ACE doesn't have to be - compiled after Version.h changes (Version.h changes every time a - new beta is produced). - -Thu Apr 9 12:27:23 1998 Carlos O'Ryan <coryan@swarm.cs.wustl.edu> - - * tests/Hash_Map_Manager_Test.cpp: POOL_SIZE was simple "const - POOL_SIZE" I changed that to "const int POOL_SIZE". Added - missing template instantiation for Static_Allocator. - -Thu Apr 09 10:50:18 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/WFMO_Reactor.cpp (dump): Fixed signed/unsigned mismatched - warnings on NT. Thanks to Darrell for noticing the bug. - -Thu Apr 9 09:37:08 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Malloc_T: Moved the ACE_Static_Allocator::dump() method into - the Malloc_T.cpp file so we'd get a vtable... Thanks to David - Levine for reporting this. - - * tests/Hash_Map_Manager_Test.cpp (run_test): Added a return - value to make G++ and David Levine happy ;-). - -Thu Apr 09 10:50:18 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/WFMO_Reactor.cpp (dump): Fixed signed/unsigned mismatched - warnings on NT. Thanks to Darrell for noticing the bug. - -Wed Apr 08 17:29:01 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/SOCK_Acceptor.cpp (open): There was a boo-boo in open() as - we were only calling bind_port() and ignoring the port supplied - by the user. Call bind_port only if port == 0. - -Wed Apr 8 01:09:52 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Malloc_T: Added a new ACE_Static_Allocator class, which is - highly optimized for allocations where (1) all the memory size - is known in advance and (2) no deletions are done. This class - will ultimately be used in TAO to improve it's startup - performance for operation tables initialization. - - * tests/Hash_Map_Manager_Test.cpp: Enhanced the test so that it - uses the new ACE_Static_Allocator feature. - - * ace/OS: Added support for the rename() method. Thanks to Ganesh - Pai for suggesting this. - - * ace/config-win32-common.h, - config-sunos*.h: - config-irix-6*.h: - Enable the ACE_HAS_WILDCARD_BIND macro. - - * ace/ACE.cpp: Modified the bind_port() method so that it is more - efficient on platforms that interpret a 0 port to bind() as a - request for the kernel to select the port. Thanks to Chris - Cleeland for pointing this out. - - * ace/Hash_Map_Manager: Made a bunch of minor changes to fix some - subtle deadlocks with intra-class locking. - - * tests/Map_Manager_Test.cpp: Restructured the tests so that we - don't keep recomputing end() and rend() of the map in each loop - iteration! - -Wed Apr 08 10:02:16 1998 David L. Levine <levine@cs.wustl.edu> - - * bin/make_release: added this script that creates ACE and/or - TAO releases. - - * Makefile: added support for make_release. See comments at - the top of the Makefile. The old release mechanism has - been retained; to use it: make release-old or - make releaseall-old. - -Wed Apr 08 00:37:33 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/ACE.cpp (bind_port): This call was not flexible enough as it - always assumed that the user wanted INADDR_ANY. Added an extra - parameter to this function where the user can pass in the addr - that she is interested in. This extra parameter defaults to - INADDR_ANY, so no existing should break. - - * ace/SOCK_Acceptor.cpp (open): Not enough information was being - passed to the new ACE::bind_port() call. Changed open() so that - the sin_addr part of the address is passed along to - ACE::bind_port(). - -Tue Apr 07 15:25:08 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-irix*.h: fixed determination of ACE_SIZEOF_LONG_DOUBLE, - depending on system release and ABI. Thanks to Robert J. - Laferriere <laferrie@gsao.med.ge.com> for figuring it all out. - -Mon Apr 06 01:29:35 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ace_ce_dll.dsp: Update project files. Rearrange dll filenames a - bit so that it's eaier to have various ACE libraries for various - CE supported CPUs. Files for every supported CPU are generated - under CPU specific directory as follow: - - CPU Directory Name - Sh3 series ace/WCE/SH aced.{lib,dll} ace.{lib,dll} - MIPS series ace/WCE/MIPS aced.{lib,dll} ace.{lib,dll} - x86 emulation ace/WCE/x86em aced.{lib,dll} ace.{lib,dll} - - * ace/config-win32-common.h: Prevent CE from using wsock32.lib. - -Sun Apr 05 16:44:31 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.h (ASYS_ONLY_MULTIBYTE_STRING): New macro. Used to - activate wchar string to char string conversion only under CE, - under other cases, the argument to this macro is treated as - a regular char string. - - * ace/INET_Addr.cpp (set): Changed to use - ASYS_ONLY_MULTIBYTE_STRING. This has gotten out of control - now. I need to think of a better name for these macros. - -Tue Apr 07 09:30:28 1998 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_lynxos.GNU: set up PATH when - cross-compiling. With this change, users don't need to - do anything special to their platform_macros.GNU or - environment to cross-compile for LynxOS (assuming that - the default platform_lynxos.GNU configuration is suitable). - -Mon Apr 06 16:28:32 1998 Steve Huston <shuston@riverace.com> - - * examples/Service_Configurator/IPC-tests/server/svc.conf: Corrected - library name to load tests from. Thanks to Sandro Doro - <XDS@necsy.it> for noticing there was a problem here. - -Sun Apr 05 10:20:26 1998 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_lynxos.GNU: only define - CROSS-COMPILE if not building on LynxOS. - - * ace/OS.h: fixed ACE_CAST_CONST definition with Sun CC. - -Sun Apr 05 00:20:52 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ACE.h: Moved "#include "ace/OS.h" out of _ACE_H_ block to - resolve the circular inclusion that occured when building ACE - Unicode Release version. - -Sat Apr 04 21:19:32 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.37, released Sat Apr 04 21:19:32 1998. - -Sat Apr 04 19:41:03 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.cpp (thr_exit): Forgot to access creation flags thru - the access function. - -Sat Apr 04 14:01:53 1998 Steve Huston <shuston@riverace.com> - - * performance-tests/Makefile: Put UDP back in dir list. - -Sat Apr 04 07:45:33 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Basic_Types.h,config-sunos4*.h,README: don't use - u_longlong_t on Suns if ACE_LACKS_U_LONGLONG_T is defined. - Define ACE_LACKS_U_LONGLONG_T in all SunOS4 config files. - Thanks to John Lindal <jafl@cco.caltech.edu> for reporting - this. - -Sat Apr 4 02:06:57 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.cpp (invoke,thr_exit): (*Win32*) We also need to consider - threads that are not spawned by ACE_Thread_Manager by checking - the validity of cached ACE_Thread_Descriptor. In any rate, - please avoid using thread functions in ACE_Thread::* and - ACE_OS::thr_* directly. - -Sat Apr 4 01:03:17 1998 Alexander Babu Arulanthu <alex@merengue.cs.wustl.edu> - - * ace/Map_Manager.cpp (map): Added map method for the iterator interface. - -Sat Apr 4 00:55:30 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * netsvcs/clients/Naming/{Client,Dump_Restore}/Makefile: - Moved all program source names to PSRC, not SRC, for - 'make depend'. - -Fri Apr 3 21:18:51 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS.cpp (cleanup): Moved ACE_Cleanup::cleanup() back into the - *.cpp file so that we have at least one non-inline virtual - function. - - * ace/OS (cleanup): Created a dummy ACE_Cleanup constructor since - the lack of this may be causing a bug with CC and the linker for - TAO's IDL compiler. Also moved the constructor and destructor of - ACE_Cleanup into the OS.i file and made them inline. We'll fix - this bug yet!!!! - -Fri Apr 03 18:45:35 1998 Steve Huston <shuston@riverace.com> - - * apps/JAWS/clients/Blobby/Makefile - * apps/JAWS/clients/Caching/Makefile - * examples/ASX/Event_Server/Event_Server/Makefile - * examples/ASX/UPIPE_Event_Server/Makefile - * examples/Logger/simple-server/Makefile - * examples/Mem_Map/IO-tests/Makefile - * examples/Reactor/Multicast/Makefile - * examples/Service_Configurator/Misc/Makefile - * examples/Shared_Malloc/Makefile - * performance-tests/Synch-Benchmarks/Makefile - Adjusted for corresponding changes to rules.bin.GNU and/or - rules.lib.GNU, below. - - * examples/Connection/non_blocking/Makefile - * examples/Service_Configurator/IPC-tests/server/Makefile - * examples/Timer_Queue/Makefile - * performance-tests/Misc/Makefile - Moved all program source names to PSRC, not SRC, for - 'make depend'. - - * performance-tests/Makefile: Removed UDP from list of dirs to build. - - * include/makeinclude/rules.bin.GNU: Set OBJEXT and VOBJS from the - SRC files in the Makefile (LSRC sets VLOBJS in rules.lib.GNU; - SRC sets VOBJS in rules.bin.GNU). - - * include/makeinclude/rules.local.GNU: depend rule will also operate - on $(PSRC) in addition to $(SRC). PSRC is new - if you have more - than one program to build in a makefile, set PSRC to the source - files for make depend. - - * include/makeinclude/rules.lib.GNU: Use VLOBJS for archive lib object - files; using VOBJS gets it confused with VOBJS in rules.bin.GNU and - makes it very difficult to build lib and exe from same Makefile. - -Fri Apr 3 10:40:10 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.cpp (invoke): We were using AfxGetThread to determine - whether a thread was created with AfxBeginThread or - _beginthreadex. This seemed to cause some subtle problem with - MSVC 4.2. Since we always cached the Thread_Descriptor in TSS - now, we can just check how a thread was created and use the - appropriate thread exiting method. Thanks to Detlef Becker - <Becker.Detlef@med.siemens.de> for tracing down this obscure - bug. - - * ace/Thread_Manager.{h,i} (flags): Added this new access function - into ACE_Thread_Descriptor to get the creation flags for a - thread. - -Fri Apr 03 10:01:32 1998 Steve Huston <shuston@riverace.com> - - * ace/ACE.cpp (handle_timed_complete): On AIX, always check a - completed non-blocking connect() because AIX sets only the write - handle from select regardless of whether the connect succeeded - or not. - - * netsvcs/clients/Naming/Dump_Restore/Dump_Restore.cpp: Removed - the #include <fstream.h> - this is taken care of via ace/OS.h. - Having the #include in Dump_Restore.cpp caused some very weird - compile errors on DG/UX. - - * netsvcs/lib/Server_Logging_Handler.cpp: Fixed template instantiations - for Base_Optimizer, non-static logging cases. - -Fri Apr 03 09:16:24 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.cpp (ACE_TSS_Info::key_in_use): return true if - thread_count_ > 0, not if it is != -1. Thanks to Chris - Lahey <CLahey@cccis.com> for finding and reporting this. - - * ace/OS.i (thr_yield): on VxWorks, changed argument of - ::taskDelay () from 0 to 1. With an argument of 0, - ::taskDelay () doesn't appear to yield the current thread. - - * tests/run_tests.vxworks: Reactor_Notify_Test is OK on - VxWorks now. - - * ace/OS.h,config-sunos5.5-sunc++-4.x.h: added ACE_CAST_CONST, - so that we can work around broken cast constness requirements - with Sun CC 4.2. It requires, for example, that a reinterpret - cast to a local pointer variable in a const member function be - const. Wrong. - - * ace/Basic_Types.i (operator/ and %): fixed comments: former - instead of latter. - -Thu Apr 02 16:23:12 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/WFMO_Reactor.i (dump): Also changed the method from HANDLE - to ACE_HANDLE. - -Thu Apr 2 15:36:20 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace: Added - - #define ACE_HAS_STL_QUEUE_CONFLICT - - to the config files which had ACE_HAS_STL_MAP_CONFLICT defined, - i.e., - config-sunos5.5-g++.h - config-sunos5.5-sunc++-4.x.h - - Thanks to Russ Noseworthy <rnosewor@objectsciences.com> for - reporting this. - -Thu Apr 2 15:14:35 1998 Sumedh Mungee <sumedh@lindy.cs.wustl.edu> - - * ace/WFMO_Reactor.h (dump): Renamed HANDLE to ACE_HANDLE. - -Thu Apr 02 14:23:11 1998 David L. Levine <levine@cs.wustl.edu> - - * tests/Reactor_Notify_Test.cpp (Supplier_Task::handle_output): - added a thr_yield () so that the Supplier_Task always gets a - chance to run. - - * tests/run_tests.sh: Reactor_Notify_Test is OK on Chorus now. - -Wed Apr 01 18:59:42 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/WFMO_Reactor.cpp (dump): Added dump() methods. - - * ace/Msg_WFMO_Reactor.cpp (wait_for_multiple_events): Reverted - Nanbor's change. The flag is properly set in - ACE_Msg_WFMO_Reactor::alertable_handle_events(). - -Wed Apr 1 15:40:01 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace: Made the WMFO_Reactor and Select_Reactor's dump() methods - virtual to underscore the fact that the Reactor_Impl's new - dump() method is also virtual. - - * ace/OS.h: Changed ACE_HAS_STL_MAP_CONFLICT to - ACE_HAS_STL_QUEUE_CONFLICT for the appropriate cases in ACE. - Thanks to Russ Noseworthy <rnosewor@objectsciences.com> for - reporting this. - - * ace/CORBA_Handler.cpp (ACE_CORBA_Handler): The reference_count_ - data member wasn't being initialized correctly. Thanks to - Mahesh Chandwani <chandwani@lucent.com> for reporting this. - - * ace: Renamed config-sunos5.5-sunc++-4.x-orbix.h to - config-sunos5.x-sunc++-4.x-orbix.h since this config file works - with SunOS 5.5 and 5.6. Thanks to Mahesh Chandwani - <chandwani@lucent.com> for reporting this. - -Wed Apr 01 16:28:30 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Reactor.{h,cpp},Reactor_Impl.h (dump): added dump () - method, which calls the (existing) dump () in the reactor - implementation. - - * ace/High_Res_Timer.* (hrtime_to_tv): fixed conversion of - ACE_hrtime_t to ACE_Time_Value to be compatible with - ACE_U_LongLong. - -Wed Apr 01 09:32:27 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Msg_WFMO_Reactor.cpp (wait_for_multiple_events): Should not - wait for all objects in this function. - -Tue Mar 31 15:27:25 1998 Steve Huston <shuston@riverace.com> - - * apps/JAWS/server/JAWS_Pipeline_Handler.cpp: Defined 'status' in - the put() function. - - * ace/config-dgux-4.x-ghs.h: Removed settings for ACE_SIZEOF_LONG_LONG - since the compiler doesn't really support a long long that ACE - is looking for (64-bit). Use the ACE-supplied ACE_ULongLong type. - -Tue Mar 31 10:46:23 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Msg_WFMO_Reactor.cpp: Changed to use - MsgWaitForMultipleObjects if we are not under NT. Also in - MsgWaitForMultipleObjectsEx, we should use NT defined flags - instead of the boolean "alertable" directly. - -Tue Mar 31 07:22:09 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.i (cond_timedwait): restored ACE_HAS_DCE_DRAFT4_THREADS - behavior to use absolute time. Special-case only on LynxOS, - which requires a relative time. Thanks to Steve Coy - <stevec@magna.com.au> for reporting this on HP-UX. - - * tests/Thread_Manager_Test.cpp (worker): cache the thread's - Signal_Catcher on entry. That way, we don't have to - dereference it in the worker's loop. It seems that a worker - thread can continue to run after main exits (and cleans up - TSS). That shouldn't happen, but it looks like it does on - Linux, LynxOS, and Digital Unix. Thanks to James CE Johnson - <jcej@lads.com> for verifying that the fix works on Linux/ - Intel and DU 4.0. - - * examples/Threads/task_five.cpp (main): wrapped return statement - at end with ACE_NOTREACHED. This will prevent compiler warnings - from Green Hills and DEC CXX, and maybe SGI cc. - -Fri Mar 27 23:04:08 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-linux-common.h: added ACE_DEFAULT_MAX_SOCKET_BUFSIZ - 65535, because some Linux 2.1.x kernels fail setsockopts if - bigger than that. Thanks to Ravi Nagabhyru <ravi@future-minds.com> - for reporting that. Also changed hard-code ACE_PAGESIZE to - ACE_HAS_GETPAGESIZE, even though getpagesize () doesn't get - used. ACE_OS::getpagesize () uses sysconf on Linux. - - * performance-tests/UDP/udp_test.cpp (main): getopt.optarg instead - of optarg, for -f option. - - * ace/IPC_SAP.cpp (enable,disable): removed the ACE_NOTREACHED - wrappers around the return statements. Seems that some versions - of Green Hills complain about an unreachable statement, but - we'll live with it. This change is only visible with SGI, - DEC CXX, and Green Hills. - - * include/makeinclude/platform_vxworks5.x_ghs.GNU: added missing - close paren after ppc conditional. Thanks to Brian Mendel for - reporting this. - -Fri Mar 27 16:47:26 1998 Steve Huston <shuston@riverace.com> - - * tests/tests.mdp: Made settings for Env_Value_Test consistent with - other tests, so that run_tests.bat can find it. - - * netsvcs/servers/servers4.mdp: Added ..\lib to Include directories - when building in Release mode. - -Fri Mar 27 13:55:13 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/High_Res_Timer.i (gettime): ACE_OS::gettimeofday doesn't - take any argument. - -Fri Mar 27 09:50:52 1998 Steve Huston <shuston@riverace.com> - - * ace/Strategies_T.cpp: Added #include "ace/Service_Repository.h" for - compile on HP. Thanks to Steve Coy <stevec@magna.com.au> for this. - -Fri Mar 27 09:21:21 1998 David L. Levine <levine@cs.wustl.edu> - - * tests/test_config.h: with Green Hills on VxWorks, rename - main to ace_main for compatibility with run_tests.vxworks. - - * ace/OS.{h,i},High_Res_Timer.* (gethrtime): added an optional - argument to allow specifying start/increment/stop to a timer. - It is currently only used on Chorus. On Chorus, using - the default ACE_HRTIMER_GETTIME will use clock_gettime () - instead of sysBench (). sysBench () disables interrupts, - so it must be used with care. - -Thu Mar 26 21:18:57 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Basic_Types.h,config-g++-common.h,config-freebsd*.h, - config-netbsd.h: moved setting of ACE_SIZEOF_LONG_DOUBLE to 12 - from Basic_Types.h to config-g++-common.h and the FreeBSD/NetBSD - config files. - -Thu Mar 26 20:53:30 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Containers.{h,i,cpp}: Changed the template parameter name - <SIZE> to <ACE_SIZE> to avoid namespace polution. Thank to John - Mulhern <9107@MN3.LAWSON.lawson.com> for suggestion this. - -Thu Mar 26 17:23:38 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.36, released Thu Mar 26 17:23:38 1998. - -Thu Mar 26 12:54:57 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/IPC_SAP.cpp (enable,disable): wrapped return statement - at end with ACE_NOTREACHED. - - * include/makeinclude/platform_vxworks5.x_ghs.GNU: added support - for x86 targets. - - * ace/Basic_Types.h: sizeof (long double) is 8 on ghs/VxWorks/i386, - not 12. - -Wed Mar 25 10:10:04 1998 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_{chorus,vxworks5.x}_ghs.GNU: - change OFLAGS to OCFLAGS. - - * include/makeinclude/platform_chorus_ghs.GNU: if debug is enabled, - link with crth_d.s.o instead of crth.s.o. - - * ace/config-g++common.h: don't #define - ACE_HAS_TEMPLATE_SPECIALIZATION with egcs 2.91. Thanks to - Sudish Joseph <sj@eng.mindspring.net> and - Loren Rittle <rittle@comm.mot.com> for reporting this - apparent bug with egcs 2.91. (egcs 2.90 builds ACE cleanly - with ACE_HAS_TEMPLATE_SPECIALIZATION.) - - * include/makeinclude/platform_vxworks5.x_g++.GNU: add VxWorks - host bin directory to PATH if its not already there. - -Tue Mar 24 19:48:44 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Proactor.cpp (timeout): Changed comparison from || to - &&. Thanks to Jason Katz <Jason@email.rts-inc.com> for reporting - this. - -Tue Mar 24 18:14:18 1998 Steve Huston <shuston@riverace.com> - - * ace/Env_Value_T.h: added #include "ace/OS.h" to get the definition - for ACE_static_cast - AIX xlC needs it for auto template generation. - -Tue Mar 24 09:00:36 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * bin/auto_compile: - Missing newline after the "DISABLED" message. - -Tue Mar 24 07:38:44 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-sunos5.5-g++.h: replaced ACE_PAGE_SIZE with - ACE_HAS_GETPAGESIZE, and added ACE_HAS_STL_MAP_CONFLICT. - Thanks to Loren Rittle <rittle@comm.mot.com> for reporting - these. - -Mon Mar 23 12:06:30 1998 David L. Levine <levine@cs.wustl.edu> - - * performance_tests/UDP/udp_test.cpp: made use of the Reactor - optional. With this change, the performance is comparable - to the original C version. - - * ace/OS.{h,i} (wslen,wscpy): added these, and WChar typedef, - for TAO. - - * include/makeinclude/platform_chorus_ghs.GNU: moved directories - to /project/doc/mvme/. - - * ace/config-vxworks5.x.h: protected #define of ACE_MAIN so that - it can be #defined on the command line. - -Mon Mar 23 11:38:43 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.h (MAXPATHLEN): Guard the MAXPATHLEN definition with - #ifndef in case if has already been defined. Thanks to "Patrick - J. McNerthney" <pat@thememedia.com> for reporting this. - -Mon Mar 23 03:18:50 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.35, released Mon Mar 23 03:18:50 1998. - -Mon Mar 23 00:03:37 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * examples/Reactor/ReactorEx/test_abandoned.cpp: Tests the - WFMO_Reactor's ability to handle abandoned mutexes. - - * examples/Reactor/ReactorEx/test_apc.cpp: Tests the - WFMO_Reactor's ability to handle regular APC notifications. - -Sun Mar 22 23:47:00 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/WFMO_Reactor.cpp: Added code to the Reactor to make it more - resilient when the wait() function returns WAIT_IO_COMPLETION - and WAIT_ABANDONED. - - ok_to_wait: Keep waiting if wait() returns WAIT_IO_COMPLETION. - - dispatch: Return <handlers_dispatched> if wait() returns - WAIT_IO_COMPLETION. - - dispatch_handles: Added code to correctly handle WAIT_ABANDONED - and to correctly figure out the index of the handler to be - dispatched. - - Thanks to Beskrovny Evgeny (evgeny_beskrovny@icomverse.com) of - Comverse Network Systems for pointing out the problem with - WAIT_IO_COMPLETION. - - * ace/Reactor.cpp (run_alertable_event_loop): Added new function - that runs the Reactor using alertable_handle_events() instead of - handle_events(). - -Sat Mar 21 19:58:42 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.i (flock_destroy): lock->lockname_ should be cast to - LPTSTR instead of char*. Thanks to Goran Lowkrantz - <Goran.Lowkrantz@infologigruppen.se> for reporting this. - - * examples/Threads/task_five.cpp: Allow users to set thread stack - size from command lines. If a user doesn't give one, it will be - set to 8k on non-Windows platforms. - -Fri Mar 20 22:07:15 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ACE_Library.{mak,mdp}: Added Msg_WFMO_Reactor. - - * ace/Asynch_IO.cpp (cancel): Added version check because MSVC 4.2 - doesn't support CancelIo. - -Fri Mar 20 21:46:16 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.34, released Fri Mar 20 21:46:16 1998. - -Fri Mar 20 20:53:27 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Parse_Node.cpp: Fixed typos. - -Fri Mar 20 14:18:53 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * performance-tests/UDP/udp_test.cpp: Completed the ACE'ification - of this test. - -Fri Mar 20 09:31:22 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: with ACE_LACKS_WCHAR_T, changed wchar_t typedef - from long to ACE_UINT32 to ensure that it is 4 bytes on - any platform. - - * performance-tests/UDP/udp_test.cpp: added this test, which - provides UDP performance statistics. Thanks to Fred Kuhns - <fredk@arl.wustl.edu> for contributing the test. - -Thu Mar 19 10:15:27 1998 Steve Huston <shuston@riverace.com> - - * ace/OS.i: Changed the dynamic library load/unload code so that - on HP-UX, g++ does the same thing as aC++ (not like HP C++), using - <dl.h> and shl_(un)load. Thanks to Kirill Rybaltchenko - <Kirill.Rybaltchenko@cern.ch> for this change. - -Thu Mar 19 10:13:35 1998 David L. Levine <levine@cs.wustl.edu> - - * performance-tests/Misc/preempt.cpp: 1) Added -f option to fork - instead of spawning new threads. 2) Added -l option for - specifying number of iterations of low priority thread/process. - 3) On Solaris, print LWP ID. - -Thu Mar 19 02:47:42 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ace_dll.dsp: Link against user32.lib for all configurations - for we now use Msg_WFMO and related APIs in ACE. (We should do - the same to ace_lib.dsp.) - - * ace/config-WinCE.h: Added dummy definition of stdin and stdout. - They don't have any real effect. - - * ace/ACE.{h,cpp} (ldopen,ldfind): Made these functions visible - under CE. Modified them to work under CE. - - * ace/OS.cpp: Added scope name to day_of_week_name and - month_name. They are used only on CE. - - * ace/OS.{h,i}: Added some dummy C methods for CE. Yacc and Lex - generated codes use these functions, although we don't really - use them in ACE. - - * ace/OS.i (exit): Added implementation of ACE_OS::exit under CE. Since - there's no ExitProcess available, I use TerminateProcess, which - is not exactly the same as using ExitProcess but that's the - closest thing I can think of. - (open): Made the char version unsupported under CE. - (fread,fwrite,...): Added dummy C methods definition for CE. - - * ace/Parse_Node.cpp: Regular CE stuff. There's a boundary - of char/wchar functions somewhere in this file. Most Svc_Conf - stuff uses char functions but the information gets converted to - wchar later. - - * ace/Service_Config.{h,i}: Added char version function for - Service_Type under CE. Service_Type (and its subclasses) are - the boundary of char/wchar functions. - - * ace/Service_Object.{h,i}: Added the chname method which always - returns char value of a Service_Object under all platforms. - - * ace/Svc_Conf.y: - * ace/Svc_Conf_y.cpp: - * ace/Object_Manager.{cpp}: Regular CE stuff. - - * ace/Svc_Conf_l.cpp: Regular CE stuff. Haven't figure out a way - to modify this from Svc_Conf.l yet. - - * ace/Service_Config.cpp - * ace/Object_Manager.{cpp}: Ifdef'ed inclusion of Naming Context - related header files out. Naming context uses Registry which - uses iostream and STL, both are not available on CE (yet?). - - * ace/WFMO_Reactor.{h,cpp}: - * ace/Proactor.{h,i,cpp}: - * ace/Asynch_IO.{h,cpp}: Ifdef'ed the body so they are not visible - under CE. CE doesn't have asynchronous IO. - -Wed Mar 18 21:30:39 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.33, released Wed Mar 18 21:30:39 1998. - -Wed Mar 18 16:36:17 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * Makefile (releaseall): There is no more TAO.tar.gz. There are - now two tar files, ACE.tar.gz and ACE+TAO.tar.gz. - ACE+TAO.tar.gz contains ALL of ACE. In particular, it will - always have the same version number and same ACE contents as - ACE.tar.gz. The intent here is that people who just want to use - ACE will use ACE.tar.gz and people who want to use ACE and TAO - will use ACE+TAO.tar.gz. Thanks to David Levine and Irfan for - helping with this. - -Wed Mar 18 15:50:30 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Log_Msg.cpp (ACE_Log_Msg): Added codes to grab - ACE_Log_Msg_Manager's lock while updating the instance_count_. - There was a race condition without the change. Thanks to Detlef - Becker's <Detlef.Becker@med.siemens.de> thread creation stress - test to reveal the bug. - -Tue Mar 17 17:20:00 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * examples/Threads/test.dsw: - * examples/Threads/task_five.cpp: - * examples/Threads/task_five.dsp: - * examples/Threads/Makefile: Added a new ACE_Task testing - program. Thanks to Detlef Becker <Detlef.Becker@med.siemens.de> - for providing the source. - -Wed Mar 18 11:00:11 1998 Alexander Babu Arulanthu <alex@merengue.cs.wustl.edu> - - * ace/Hash_Map_Manager.cpp (map): This method has been added to - the ACE_Hash_Map_Iterator_Base class. - -Wed Mar 18 09:25:30 1998 Chris Gill <cdgill@tango.cs.wustl.edu> - - * ace/ace.dsw: added user32.lib to link for dynamic library on NT - for all builds. - -Wed Mar 18 09:22:13 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Containers.cpp,Hash_Map_Manager.cpp,Map_Manager.cpp - (operator *): initialize retv to 0 to avoid warning about - using it before setting its value. - - * tests/Thread_Mutex_Test.cpp (test): added no-op statement to - suppress GHS warning about unused local variable "mutex". - - * tests/Priority_Buffer_Test.cpp (consumer): added no-op statement - to suppress GHS warning about unused local variable "cur_priority". - - * ace/README: removed ACE_HAS_BOOL. It was unused. - - * performance-tests/Misc/basic_func.{h,cpp},basic_perf.cpp: - With MSVC 5.0, do something in the empty iteration because - otherwise it gets completely optimized away. Thanks to - Alexander Ovsiankin <sasha@technologist.com> for providing - this fix. - -Tue Mar 17 23:53:56 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * examples/Threads/task_five.cpp (main): Reformatted Detlef's new - torture test so that it conforms to ACE programming conventions. - -Tue Mar 17 23:22:20 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Reactor.cpp (ACE_Reactor): If - ACE_USE_MSG_WFMO_REACTOR_FOR_REACTOR_IMPL is defined, use - ACE_Msg_WFMO_Reactor for ACE_Reactor::instance(). - - * examples/Reactor/ReactorEx/test_window_messages.cpp: Tests - Msg_WFMO_Reactor's ability to handle regular events and window - messages. - -Tue Mar 17 21:41:57 1998 David L. Levine <levine@cs.wustl.edu> - - * examples/Threads/task_five.cpp (svc): use ACE_OS::sleepq - instead of ::Sleep. (main): declare "ii" outside of the - first loop because it is used in other loops. - -Tue Mar 17 16:02:26 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/WFMO_Reactor: Changes were made to WFMO_Reactor to make it - easier to extend it to handle window messages: - - - dispatch_handler() checks if there are window messages that - need to be dispatched if the <index> matches <max_handlep1>. - If there are, they are dispatched by calling - dispatch_window_messages(). - - - dispatch_window_messages() is a noop for WFMO_Reactor. - - - Checking the return values from Wait*() functions was changed - as (WAIT_OBJECT_0 + nCount) is now a valid index. - - - Checking for remaining handles was abstracted into - poll_remaining_handles(). - - - Expiring of timer was abstracted out into expire_timers(). - Only the owner thread expires the timers. - - - calculate_timeout() was removed from - wait_for_multiple_events(). - - * ace/Msg_WFMO_Reactor: A new Reactor that specializes the - WFMO_Reactor. WFMO_Reactor uses WaitForMultipleObjects while - Msg_WFMO_Reactor uses MsgWaitForMultipleObjects. This allows - the Msg_WFMO_Reactor to react to Windows and COM/DCOM messages - in addition to the regular WFMO_Reactor events. - - Thanks to Beskrovny Evgeny (evgeny_beskrovny@icomverse.com) of - Comverse Network Systems for contributing this to ACE. - -Tue Mar 17 13:15:11 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Future.cpp: Added missing ACE_MT macros so that we now - compile correctly if we don't have threads. Thanks to Laura - Paterno <lpaterno@d0chb.fnal.gov> for reporting this. - -Tue Mar 17 15:13:17 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/High_Res_Timer.cpp (dump): fleshed out the implementation. - - * ace/Profile_Timer.{h,cpp} (dump): implemented for platforms - w/o ACE_HAS_PRUSAGE_T and ACE_HAS_GETRUSAGE. - - * ace/OS.i (gethrtime): with ACE_HAS_CLOCK_GETTIME, carefully - construct return value to avoid arithmetic overflow if - ACE_LACKS_LONGLONG_T. - -Mon Mar 16 22:18:07 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/IPC_SAP.cpp: Moved return values outside of the #ifdefs to - quiet warning messages with KAI C++. Thanks to Laura Paterno - <lpaterno@d0chb.fnal.gov> for reporting this. - -Tue Mar 17 00:04:53 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Svc_Conf.y: - * ace/Svc_Conf_y.cpp: Applied "Regular CE Tricks [TM]". - -Mon Mar 16 23:34:39 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/README: - * ace/config-WinCE.h: Added ACE_LACKS_STRRCHR. - - * ace/OS.i: Added implementation of ACE_OS::strrchr () when - platforms don't provide one. - -Mon Mar 16 16:08:45 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/High_Res_Timer.cpp: set global_scale_factor_ to 1000 with - ACE_HAS_CLOCK_GETTIME, because clock_gettime has nanosecond - resolution, and with PSOS. - - * ace/Profile_Timer.{h,cpp},tests/Handle_Set_Test.cpp: use units - of microseconds instead of seconds if ACE_LACKS_FLOATING_POINT. - - * ace/OS.i (hostname): on Chorus, try the HOST environment variable - if ::gethostname () returns an empty string. - -Mon Mar 16 13:04:58 1998 Seth Benjamin Widoff <sbw1@waltz.cs.wustl.edu> - - * ace/OS.h: - Extended the ACE_HAS_STL_MAP_CONFLICT to include 'queue' as well - as 'map.' 'queue' is defined in sys/stream.h, which is included in - OS.h by netinet/in.h and sys/timod.h. - -Mon Mar 16 00:22:40 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * CE-status: Updated CE port status. - - * ace/config-WinCE.h: Added macro definitions for EOF and BUFSIZ. - - * ace/OS.{h,i} (ace_isspace): Added both char and wchar version - functions. Notice that this can't be called isspace because - most compilers implement isspace as a macro. - (dlerror,dlsym): Added wchar version of these functions under CE. - - * ace/OS.cpp (fork_exec): Modified parameters passed to - CreateProcess under CE. - - * ace/ARGV.{h,i,cpp}: Unicodize this class under CE. Notice that - under CE, you won't be able to use environment variable - substitution since there's no environment variable at all. - - * ace/Reactor.cpp: Don't include Service_Config.h and - WFMO_Reactor.h under CE for now. Prevent Reactor from - reconfigure the service configurator. - - * ace/Task.cpp: Don't include Service_Config.h under CE for now. - Prevent Task_Base from reconfigure the service configurator. - - * ace/SString.cpp: Don't include Service_Config.h under CE for - now. - - * ace/Svc_Conf.h: - * ace/Shared_Object.{h,cpp}: - * ace/Service_Type.{h,i,cpp}: - * ace/Service_Repository.{h,cpp}: - * ace/Service_Object.{h,i,cpp}: - * ace/Service_Manager.{h,cpp}: - * ace/Service_Config.{h,i,cpp}: Unicodize service configurator - related classes. I still need to find a point which breaks the - usage of char functions that flex and yacc use and wchar - functions that most CE functions use. - - * ace/Get_Opt.{h,cpp}: - * ace/Local_Token.{h,i,cpp}: - * ace/Map_Manager.cpp: - * ace/Message_Queue.cpp: - * ace/Module.{h,i,cpp}: - * ace/Parse_Node.cpp: - * ace/Pipe.cpp: - * ace/Stream.{h,cpp}: - * ace/Stream_Module.{h,cpp}: - * ace/Token_Manager.{h,cpp}: - * ace/Activation_Queue.cpp: Unicodize classes. Regular CE stuff. - -Sat Mar 14 06:43:32 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Basic_Types.h,config-win32-common.h,config-irix*.h: - moved WIN32 and SGI-specific code out of Basic_Types.h and - into their respective config files. - - * ace/Object_Manager.h: changed ACE_HAS_NON_STATIC_OBJECT_MANAGER - to ACE_HAS_NONSTATIC_OBJECT_MANAGER in comments. Thanks to - Kirill Rybaltchenko <Kirill.Rybaltchenko@cern.ch> for reporting - this. - - * include/makeinclude/platform_chorus_ghs.GNU: set PATH, and - remove /tmp/gh_*.[os] files that GreenHills leaves. - - * ace/INET_ADDR.cpp (get_host_name): on Chorus, save errno in - error so that it gets propagated back to the caller. - -Sat Mar 14 00:13:22 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Task.h (ACE_Task_Base): Give a default value of 0 for the - Thread Manager so we can use this as a base class. - - * tests/Future_Test.cpp: Since the Scheduler provides its own - activation queue it doesn't make sense to inherit from ACE_Task - (which provides its own queue) but instead ACE_Task_Base - (which doesn't define its own queue). Thanks to Loren Rittle - <rittle@comm.mot.com> for reporting this. - - * examples/Threads/future[12].cpp: Since the Scheduler provides - its own activation queue it doesn't make sense to inherit from - ACE_Task (which provides its own queue) but instead - ACE_Task_Base (which doesn't define its own queue). Thanks to - Loren Rittle <rittle@comm.mot.com> for reporting this. - - * examples/Threads/future[12].cpp: Removed the explicit call to - scheduler_->close() to avoid closing the scheduler down twice. - Thanks to Loren Rittle <rittle@comm.mot.com> for reporting this. - -Fri Mar 13 23:41:44 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Process.cpp (ACE_Process_Options): Moved column out of - ifdef'ed area. - -Fri Mar 13 21:17:51 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Process.cpp (ACE_Process_Options): reordered initializers - to match declaration order. - -Fri Mar 13 17:30:16 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Process.{h,i,cpp}: As CE's CreateProcess is basically - broken, you can only start a new process using ACE_Process. You - cannot set any of environment, process attributes, thread - attributes, current directory, stratup info, and security - attributes. I have modified these files so that most attribute - access functions just do nothing. Some of "do nothing" - functions were defined as ACE_INLINE on CE. - - * ace/Memory_Pool.cpp: Regular CE stuff. - -Fri Mar 13 17:42:17 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/WFMO_Reactor.cpp (handle_signal): If there are still - notification in the queue, we need to wake up again. - -Fri Mar 13 09:09:08 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Profile_Timer.*,config-chorus.GNU,README, - tests/Handle_Set_Test.cpp (test_performance): added - ACE_LACKS_FLOATING_POINT support. - - * include/makeinclude/platform_chorus_ghs.GNU: removed - -X445 from CFLAGS. I don't know what it does. - -Fri Mar 13 01:10:36 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/WFMO_Reactor: The following changes were made to make the - registration and removal of event_handlers from WFMO_Reactor - more robust. Thanks to Jonathan Biggar <jon@floorboard.com> for - helping out. - - - ACE_WFMO_Reactor_Handler_Repository::Common_Info now contains - the delete_entry_ and close_masks_ fields as these were now - common to all three structures. - - - handle_additions() and handle_deletions() were changed to - make_changes_in_current_infos(), - make_changes_in_suspension_infos(), and - make_changes_in_to_be_added_infos() - - - Correct range for suspend_handlers() is 0 to max_handlep1_ - rather than (max_handlep1_ - 1) - - - suspend_handlers(), - resume_handlers(), - add_network_events_i(), - unbind_i(), - suspend_handler_i(), - resume_handler_i(), - unbind_all(), - - now go through the to_be_add_info_ array when making changes. - - * examples/Reactor/ReactorEx/test_prerun_state_changes.cpp: New - test added to check the Reactor's ability to handle state - changes before getting a chance to run. - - * examples/Reactor/ReactorEx/test_talker.cpp (main): Remove the - Procator from the Reactor so that the Proactor is not called - after it has been destroyed. - -Thu Mar 12 19:36:43 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ACE-INSTALL.html: Tidied up a bunch of things in this file. - Also, created a new $ACE_ROOT/os-patches/ directory and moved - the patches for Linux and LynxOS into this in order to - consolidate the patches into one location. - - * ace/WFMO_Reactor.cpp (upcall): Cleaned up a bunch of code to (1) - avoid a nasty "dangling else" bug and (2) simplify a bunch of - if/if statements. Thanks to Jonathan Biggar - <jon@floorboard.com> for pointing this out. - -Thu Mar 12 18:21:15 1998 James C Hu <jxh@cs.wustl.edu> - - * ace/config-linux-common.h: Incorporated changes that were - suggested by Robert Hanzlik <robi@codalan.cz> to get ACE to - compile on Linux using glibc 2.1 and libg++/gcc 2.8. - -Thu Mar 12 17:25:28 1998 James C Hu <jxh@cs.wustl.edu> - - * ACE-INSTALL.html: Added a table to Supported Platforms and - Compilers. - -Thu Mar 12 16:28:32 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * Apply regular string manipulation for CE to following files: - ACE.cpp ARGV.cpp Acceptor.cpp Activation_Queue.cpp - Asynch_Acceptor.cpp Asynch_IO.cpp CORBA_Handler.cpp - Connector.cpp Containers.cpp DEV_Addr.cpp Dynamic_Service.cpp - FIFO.cpp FIFO_Recv.cpp FIFO_Recv_Msg.cpp FIFO_Send.cpp - FIFO_Send_Msg.cpp FILE_Addr.cpp FILE_Connector.cpp Filecache.cpp - Future.cpp Get_Opt.cpp Hash_Map_Manager.cpp LSOCK.cpp - LSOCK_CODgram.cpp LSOCK_Connector.cpp LSOCK_Dgram.cpp - Local_Name_Space_T.cpp Local_Tokens.cpp Local_Tokens.i - Malloc_T.cpp Map_Manager.cpp Mem_Map.cpp Memory_Pool.cpp - Message_Queue.cpp Module.cpp Name_Proxy.cpp - Name_Request_Reply.cpp Naming_Context.cpp Obstack.cpp - Parse_Node.cpp Pipe.cpp Priority_Reactor.cpp Proactor.cpp - Process_Manager.cpp Read_Buffer.cpp Registry_Name_Space.cpp - Remote_Name_Space.cpp Remote_Tokens.cpp SPIPE_Acceptor.cpp - SV_Semaphore_Complex.cpp SV_Semaphore_Simple.cpp - Select_Reactor.cpp Service_Config.cpp Service_Main.cpp - Service_Manager.cpp Service_Repository.cpp Service_Types.cpp - Singleton.cpp Strategies_T.cpp Stream.cpp Svc_Handler.cpp - TLI.cpp TLI_Acceptor.cpp TLI_Connector.i Task_T.cpp - Time_Request_Reply.cpp Timer_Queue_Adapters.cpp - Token_Collection.cpp Token_Invariants.cpp Token_Manager.cpp - Token_Request_Reply.cpp WFMO_Reactor.cpp WFMO_Reactor.i - XtReactor.cpp. - -Thu Mar 12 15:37:14 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Read_Buffer.* (read): null terminate the result buffer. - Also, changed "inline" to "ACE_INLINE" and updated #includes - so that ACE inlining works. - -Thu Mar 12 02:19:53 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/WFMO_Reactor.cpp (upcall): In the Select_Reactor, when it - upcalls to an event_handler which returns a -1, the - Select_Reactor only clears the specific mask bit for the - handler. For example, for a call to - Event_Handler::handle_input() which returns -1, the - Select_Reactor clears the READ_MASK bit. - - In the WFMO_Reactor, if any handler returns a -1, then it clears - ALL of the mask bits for the event handler. - - So, fixed WFMO_Reactor so that it works like Select_Reactor in - this regard. - - Thanks to Jonathan Biggar <jon@floorboard.com> for pointing this - discrepancy out. - - * ace/WFMO_Reactor.cpp (dispatch_handles): The correct range check is - {WAIT_OBJECT_0,WAIT_OBJECT_0 + this->handler_rep_.max_handlep1 () - index} - and not - {WAIT_OBJECT_0,WAIT_OBJECT_0 + this->handler_rep_.max_handlep1 ()} - - Thanks to Evgeny Beskrovny evgeny_beskrovny@icomverse.com for - pointing this out. - -Wed Mar 11 15:20:09 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS.i (thr_getspecific): Added comments to the code to - explain the unusual step of saving errno across calls to - TlsGetValue() on Win32. Thanks to Tim Harrison for clarifying - this. - -Wed Mar 11 00:39:29 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * CE-status: A temporary file to trace the progress of the CE - port. - - * ace/ace_ce_dll.dsp: Added more .cpp files. - - * ace/High_Res_Timer.{h,cpp}: Undefined functions print_ave and - print_total. These two function use STDOUT which is not - supported on CE. We should probably modify the functions so - they write results to Log_Msg. - (get_register_scale_factor): This function always returns 1 on - CE to prevent the use of High_Res_Timer. - (get_env_global_scale_factor): Always returns -1 on CE because - environment variables are not supported. - - * ace/Thread_Manager.cpp (spawn_i): WinCE doesn't support - DuplicateHandle, therefore the best we can do now is just pass - back the handle to the spawned thread. Notice that you *can't* - close the handle by yourself under CE. Perhaps we shouldn't - even try to pass this info back. ??? - Applied regular CE porting tricks. - - * ace/SOCK_Dgram_Mcast.{h,cpp}: - * ace/SOCK_Dgram_Bcast.{h,cpp}: Changed some function signatures - to use wchar on CE. Applied regular CE porting tricks. - - * ace/IO_SAP.cpp: - * ace/SOCK_CODgram.cpp: - * ace/SOCK_Dgram.cpp: - * ace/Select_Reactor.cpp: - * ace/Timer_{Hash,Heap,List,Queue,Wheel}_T.cpp: Applied regular CE - porting tricks. Erm, I'm running out of tricks. - -Tue Mar 10 20:27:57 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Reactor.h (reset_event_loop): Made the signature match with - its definition in .cpp file. - -Tue Mar 10 15:07:00 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Svc_Handler.cpp, - ace/Module.cpp: - Fixed some problems where Svc_Handlers in Modules were being - deleted prematurely. Thanks to Steve Coy <stevec@magna.com.au> - for reporting this. - - * ace/Reactor.h (ACE_Reactor): Added a "reset_event_loop" method - so that it's possible to reset the loop again once the - end_event_loop() method is called. Thanks to Eric Hopper - <ehopper@globalmt.com> for reporting this. - - * ace/IOStream_T.cpp (ACE_IOStream): Moved some code to work - around some bugs with MSVC++ 5.0. Thanks to James CE Johnson - <ace-users@lads.com> for reporting this. - -Tue Mar 10 15:57:54 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Handle_Set.cpp (dump): Cast mask_.fd_count to size_t. - Depend on whether you are using standard C++ library, the type - of fd_count count be size_t or ssize_t. Thanks to Darrell for - discovering this. - -Tue Mar 10 13:22:24 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Asynch_IO.cpp (cancel): CancelIo is not supported on Win95. - - * ace/SString.cpp (set): Use ACE_WString::wstrlen(). - -Tue Mar 10 06:59:42 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/SString.cpp (WString::set): added support for platforms - that don't have wchar_t, or that treat wchar_t and ACE_USHORT16 - as different types. - - * ace/OS.h,Basic_Types.h: moved ACE_USHORT16 typedef from OS.h to - Basic_Types.h. - - * ace/High_Res_Timer.h: fixed comment: timer.start () instead of - timer.stop () (twice). - -Mon Mar 09 22:15:43 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/SString.cpp (ACE_CString::operator=): Special cases have to - considered carefully so that memory is managed properly. This - is specially true when the ACE_CString::NULL_CString_ is used. - - * ace/SString.h (ACE_CString and ACE_WString): Added set methods - that allows the user to set the strings without having to use - the assignment operator. - -Mon Mar 09 16:11:16 1998 David L. Levine <levine@cs.wustl.edu> - - * examples/Map_Manager/test_hash_map_manager.cpp: added #include - of ace/SString.h and template instantiations. Replaced gets - with ACE_OS::fgets. - - * examples/Makefile,examples/Map_Manager/Makefile: added - Map_Manager to examples DIRS and added RCS keyword string. - -Mon Mar 9 14:59:20 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * include/makeinclude/platform_irix6.x-sgic++.GNU: - Added the definition for MATHLIB. - -Mon Mar 09 01:23:49 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ace_ce.dsw: - * ace/ace_ce_dll.dsp: New project/workspace files for WinCE. This - is a temporary workspace only and should be merged into ACE mail - Win32 project/workspace when the port is stable enough. - - * ace/config-WinCE.h: Added ACE_LACKS_GETS. - - * ace/Synch.i (get_nesting_level): Defined it as not supported on - CE. CE's CriticalSections don't have RecursionCount defined. - We should be able to simulate this, however. - - * ace/Process.{h,cpp}: - * ace/SString.{h,cpp}: Moved class ACE_Tokenizer from Process.* to - SString.*. Although ACE_Tokenizer is used in - ACE_Process[_Options], it really is a more general utility - class. - (operator<<): Conditionally compile these functions only when we - are not running ACE on CE. - - * ace/OS.i (atoi): Added inline implementation for wchar version. - (gets): Remove condition !defined (ACE_HAS_WINCE). This should - be taken care of by ACE_LACKS_GETS. - - * ace/OS.h (atoi): Added wchar version of this function. - (ASYS_WIDE_STRING): This macro converts char strings to - wchar strings when ACE_HAS_MOSTLY_UNICODE_APIS is defined. - (ASYS_MULTIBYTE_STRING): This macro converts wchar strings back - to char strings *only* when ACE_HAS_MOSTLY_UNICODE_APIS is - defined. - - * ace/INET_Addr.{h,cpp}: Redefined interface signatures on WinCE - using ASYS_* macros. The principle here is to pass wide char - strings in and get char strings out. That's because we will use - mostly wchar strings in CE but most winsock APIs return (or uses - char strings.) Doing so should reduce conversion between them - to minimum. - Converting debug messages with ASYS_TEXT. - - * ace/Log_Msg.cpp (open): - * ace/ACE.cpp (get_ip_interfaces): Only convert the wide character - string to a char string when ACE_HAS_MOSTLY_UNICODE_APIS flag is - not set using the ASYS_MULTIBYTE_STRING macro. - - * ace/SOCK.cpp: - * ace/SOCK_Acceptor.cpp: - * ace/Malloc.cpp: - * ace/Malloc_T.cpp: - * ace/Synch.cpp: - * ace/Message_Block.cpp: - * ace/IPC_SAP.cpp: - * ace/Handle_Set.cpp: - * ace/Addr.cpp: Converting debugging messages to wchar strings for - CE using the ASYS_TEXT macro. - -Sun Mar 08 14:29:30 1998 David L. Levine <levine@cs.wustl.edu> - - * apps/Gateway/Gateway/Event_Channel.cpp (initiate_acceptors): - changed SUPPLIER_CONNECTOR to SUPPLIER_ACCEPTOR. Thanks - to Dani Flexer <danif@ivory-sw.com> for reporting this. - - * ace/OS.i (cond_timedwait): removed "struct" from declaration - of local timespec_t relative_time, because timespec_t can - be a typedef. Thanks to Steve Coy <stevec@magna.com.au> for - reporting this. - - * Makefile: tag from ACE_wrappers, not its parent directory, - to avoid confusing CVS. - -Sat Mar 07 16:46:58 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE version 4.4.32, released Sat Mar 07 16:46:58 1998. - -Fri Mar 06 08:37:44 1998 David L. Levine <levine@cs.wustl.edu> - - * bin/ace_ld: made perl4-compatible. Thanks to - Jerry Bickle <glbick@most.fw.hac.com> for reporting this. - - * ace/os.i: (sigwait): removed "::" from before sigwait () call - with ACE_HAS_ONEARG_SIGWAIT, in case it's a macro. Thanks - to Kirill Rybaltchenko <Kirill.Rybaltchenko@cern.ch> for - reporting this. - - * bin/create_ace_build: made perl4-compatible. Thanks to - Carlos for noticing this. - - * include/makeinclude/platform_vxworks5.x_g++.GNU: added - WIN32 host support. Thanks to Paul von Behren - <Paul_von_Behren@stortek.com> for providing the patches. - - * ACE-INSTALL.html: added information from Paul von Behren - <Paul_von_Behren@stortek.com> on building on NT hosts for - VxWorks targets. - - * tests/Recursive_Mutex_Test.cpp: 50 iterations instead of 100 - on Chorus. - - * include/makeinclude/platform_chorus_ghs.GNU: added -G to - LDFLAGS. - - * tests/run_tests.sh: added capability for executing the tests - on a remote (Chorus) target. - - * examples/System_V_IPC/Makefile: added clean and realclean targets - for VXWORKS. - - * Makefile: cd .. before tagging. - -Thu Mar 05 12:36:52 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.h (ACE_TRACE): Added a missing right paren. - -Thu Mar 05 10:04:06 1998 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_chorus_ghs.GNU: this works, - though it may not be optimum. - -Thu Mar 05 01:33:08 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/config-WinCE.h: Added more CE stuff. Ostream is now - redefined as FILE on CE. - - * ace/OS.{i,cpp}: - * ace/Synch_T.cpp - * ace/Message_Queue.cpp: - * ace/SOCK_Connector.i: Added ASYS_TEXT to numerous - ACE_DEBUG/ACE_ERROR macros. We must use wide char strings in - CE. - - * ace/ACE.{h,cpp}: Redefined sock_error to return wchar string on - CE. Added ASYS_TEXT to debugging messages. - - * ace/Trace.{h,cpp}: - * ace/Log_Msg.{h,cpp}: Made it UNICODE compatible. - - * ace/Log_Record.cpp: Changed how ostream is used on CE. Although - the code here become some how platform dependent, fortunately, - it can be shielded out from the users thru various macros. - Added ASYS_TEXT. - - * ace/SString.h: Undefined friend relations of ACE_?String with - ostream on CE. CE doesn't support <iostream>. - - * ace/OS.h: Changed the definition of ASYS_TEXT to use _TEXT. - This macro can be used with other compiler defined macros - (e.g. __FILE__.) Added ASYS_TEXT to adorn parameters of - ACE_TRACE so we don't need to modify our codes to make it - portable to CE. - - * ace/Log_Msg.cpp: Removed macro definition of ACE_WSPRINTF. It - didn't appear to be used anywhere. - -Wed Mar 04 09:32:32 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.cpp (readv,writev): added static cast of iov_len to int, - to avoid unsigned comparison < 0. - - * ace/Containers.cpp (ACE_Fixed_Set_Iterator::done), - ace/Map_Manager.cpp (ACE_Map_Iterator_Base::done): - removed const from static cast to avoid warnings about - meaningless type qualifier from GreenHills. - - * ace/Basic_Types.i (ACE_U_LongLong): added missing ACE_INLINE's. - - * ace/Svc_Conf.y,Svc_Conf_y.cpp, etc/Svc_Conf_y.cpp.diff - (svc_location): delete svc_initializer after initializing - svc_location. Thanks to Tom Arbuckle - <arbuckle@uran.informatik.uni-bonn.de> for reporting this. - - * include/makeinclude/platform_linux_lxpthreads.GNU: separated - SOBUILD into compile and link steps because the - Service_Configurator/server/* examples are unable to - use files compiled with -fPIC and just labeled as .so as shared - ELF objects. Thanks to Tom Arbuckle - <arbuckle@uran.informatik.uni-bonn.de> for reporting this. - - * include/makeinclude/platform_chorus_ghs.GNU: explicitly link - in ioassign.o to resolve _cerr. - - * tests//SOCK_Connector_Test.cpp, - ace/config-{chorus,lynxos,netbsd,psosim-g++,vxworks5.x, - win32-common}.h: - added ACE_LACKS_GETHOSTENT, so that SOCK_Connector_Test doesn't - have to enumerate the platforms that don't have ::gethostent (). - - * tests/Basic_Types_Test.cpp (main): if ACE_LACKS_LONGLONG_T, use - ACE_U_LongLong instead of long long. - -Tue Mar 03 22:15:16 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Remote_Name_Space.cpp: Replaced a bunch of usages of - ACE_WString::rep (). They were passed to ACE_Name_Request which - creates another copy for itself and nobody ever cleans up these - memories. Thanks to John Mulhern <9107@MN3.LAWSON.lawson.com> - for reporting this. - - * ace/OS.h (THR_DETACHED): Defined this macro so that we can - simulate detached threads on NT (Hi Irfan! ;-). Thanks to Andre - Arpin <arpin@adan.kingston.net> for prividing the use case of - detached threads. - - * ace/Thread_Manager.cpp (exit): Added detached threads - manipulating codes here. We now can simulate behaviros of UNIX - detached threads on NT (no one cares about their return status.) - -Tue Mar 03 15:05:42 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Basic_Types.{h,i} (ACE_U_LongLong::operator *=): updated - with James Hu's implementation. Now it's fast :-) - - * tests/Time_Value_Test.cpp (test_ace_u_longlong): added another - test of operator *=. - -Mon Mar 2 19:21:36 1998 Douglas C. Schmidt <schmidt@waltz.cs.wustl.edu> - - * ace/OS.h: Changed all uses of "/localname" and "/globalname" to - "localname" and "globalname" to avoid problems on UNIX. Thanks - to John Mulhern <9107@MN3.LAWSON.lawson.com> for reporting this. - - * ace/OS.h: Changed "C:\\temp\\" to "C:\\temp" to avoid some minor - problems on NT. Thanks to John Mulhern - <9107@MN3.LAWSON.lawson.com> for reporting this. - - * netsvcs/lib/Log_Message_Receiver.cpp - (Log_Message_Receiver_Impl): Removed some stray inline keywords. - Thanks to Russ Noseworthy <rnosewor@objectsciences.com> for - reporting this. - - * ace/OS.i (sigwait): Added support for POSIX sigwait(). Thanks - to Russ Noseworthy <rnosewor@objectsciences.com> for reporting - this. - - * ace/Containers.cpp (done): - ace/Map_Manager.cpp (done): Corrected some misplaced "consts" in - the ACE_static_cast macro. Thanks to Russ Noseworthy - <rnosewor@objectsciences.com> for reporting this. - - * include/makeinclude/platform_sunos5_sunc++.GNU (CXX): Commented - out the -pta option since it might not be necessary. If this - breaks code then we'll need to reevaluate this change. Thanks - to Russ Noseworthy <rnosewor@objectsciences.com> for suggesting - this. - -Mon Mar 02 17:06:46 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Thread_Manager.cpp (exit): Added Win32 thread exit - functionality. When a thread function falls out of scope, this - functions is called by ACE_Thread_Control::exit and it will not - call ACE_OS::thr_exit from here. Under this scenario, the - ACE_Thread_Manager::exit simply falls thru the end. If a user - call this function directly, this function will delete - ACE_Thread_Exit::instance() first, which simulate the cleanup - routine executed when a thread fall out of end by calling - ACE_Thread_Manager::exit recursively. After this is done - (returned from the second call,) this function will then call - ACE_OS::exit, which clean up the TSS objects and then really - exit the thread. Thanks to Gordon Watts <gwatts@FNAL.GOV> for - pointing this out. - - * ace/OS.cpp (thr_exit): Added Win32 thread exit functionality. - ACE_OS::thr_exit now will try to clean up TSS objects by calling - ACE_OS::cleanup_tss. You really should _not_ calling this - function directly. Esp. from the main thread because I don't - know how to tell if you are calling it from main thread thread - or not and thus memory leak might occur. - -Mon Mar 02 09:13:26 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-sunos5.5-sunc++-4.x.h: replaced ACE_PAGE_SIZE with - ACE_HAS_GETPAGESIZE, and added ACE_HAS_STL_MAP_CONFLICT. - Thanks to J. Russell Noseworthy <rnosewor@objectsciences.com> - for these recommendations. - - * ace/config-sunos5.6-sunc++-4.x.h: just #include - config-sunos5.5-sunc++-4.x.h. There's no difference between - the two at this point. - - * include/makeinclude/platform_sunos5_sunc++.GNU: added -R./ and - -mt to LDFLAGS. Added -fast to LDFLAGS if "fast" flag is - enabled. Thanks to J. Russell Noseworthy - <rnosewor@objectsciences.com> for these recommendations. - - * include/makeinclude/platform_sunos5.6_sparc_sunc++.GNU: just - include platform_sunos5_sunc++.GNU. There's no difference between - the two at this point. There is one significant net change to - the 5.6 platform: -s is no longer included in LDFLAGS. - - * ace/config-chorus.h: added ACE_LACKS_ACE_IOSTREAM with ghs, because - its iostream doesn't have separate operators for signed/unsigned - char. - -Sun Mar 1 13:54:02 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * Makefile: Added cvs commands to automatically tag the CVS - repository with version number 'ACE_9_9_99' tag and 'Current' - tag. - -Sun Mar 1 12:26:13 1998 Alexander Babu Arulanthu <alex@merengue.cs.wustl.edu> - - * ace/Hash_Map_Manager.cpp (reset_i): - * ace/Hash_Map_Manager.h: - Added reset () method for the Hash_Map_Iterator. - -Sun Mar 01 10:59:27 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Asynch_IO.cpp (cancel): Changed CancelIO to CancelIo. - - * ace/Synch.h (ACE_Lock): Added a virtual destructor to ACE_Lock. - -Sun Mar 01 08:38:00 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: use the new ACE_DEFAULT_SYNCH_TYPE for the default - mutex, etc., type, so that it can be set VxWorks. The ACE default - type on VxWorks is SEM_Q_FIFO, the same as VxWorks' default. - The ACE default can be easily overridden in config.h by - defining ACE_DEFAULT_SYNCH_TYPE. - - * ace/OS.i (sema_init): on VxWorks, use the type argument for the - sema type, instead of hard-coding to SEM_Q_FIFO. - - * ACE-INSTALL.html: updated Linux/alpha "as" patch. - -Sat Feb 28 12:20:57 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Message_Queue.cpp: Fixed the implementation of - wait_not_{full,empty}_cond() so that they utilize the - ACE_Time_Value it's passed. In addition, revised the - implementation of peek_dequeue_head() so that it uses the new - wait_not_empty_cond() implementation. Thanks to Tom Venturella - <tventurell@mdc.com> for reporting this. - -Fri Feb 27 15:47:15 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE-INSTALL.html: added very useful information from - Chris Ryan <cryan@qualcomm.com> about build ACE with - NT/Tornado for VxWorks. - - * ace/config-chorus.h: use ACE_NEEDS_READV/WRITEV with GHS as well - as with g++. - - * ace/Basic_Types.{h,i} (ACE_U_LongLong): added a bunch of arithmetic - operators for Carlos :-) The operator *= is _very_ slow, until we - come up with the right way to do it. - - * tests/Time_Value_Test.cpp (test_ace_u_longlong): added some tests - of the new operators. - -Fri Feb 27 13:42:09 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/ACE.cpp (recv_n): Removed the MSG_PEEK flag since it didn't - seem to work correctly. - -Fri Feb 27 11:18:28 1998 Chris Cleeland <cleeland@lambada.cs.wustl.edu> - - * ace/Log_Msg.cpp: Added three new conversion operators: "%{" - calls inc() to increase the trace_depth, "%}" calls dec() to - decrease the trace_depth, and "%$" is used in place of "\n" to - insert a newline followed by appropriate current indentation. - -Thu Feb 26 21:36:27 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.31, released Thu Feb 26 21:36:27 1998. - -Thu Feb 26 16:48:53 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/OS.h: Added new macros for ACE_FILE_STREAM and - ACE_FILE_CONNECTOR to be consistent with the ACE_SOCK_* stuff, - etc. Thanks to Ganesh Pai <gpai@voicetek.com> for reporting - this. - -Thu Feb 26 11:14:21 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Parse_Node.cpp (ACE_Location_Node): reordered initializers - to match declaration order. - -Thu Feb 26 02:42:57 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Service_Config.cpp (load_static_svcs): Removed explicit - constness cast from the allocation function. - -Wed Feb 25 23:08:44 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * tests/MT_Reactor_Timer_Test.cpp: Made this test compile on - platforms that lack threads and also updated the programming - style to be consistent with other parts of ACE. - - * tests/Priority_Task_Test.cpp: Made this test compile on - platforms that lack threads. - - * ace/Svc_Conf.h: Fixed the prototype for - ace_create_service_type() to reflect the elimination of "const - void *". - - * ace/Parse_Node: Eliminated more uses of "const void *" and - changed them to plain ol' "void *". - - * ace/Service_Types.cpp (fini): Changed the "const void *" to be - simply "void *" since it didn't seem to make any sense they way - it was! Thanks to Stephen Coy <stevec@magna.com.au> for - insisting on this... - - * ace/config-sunos5.*.h: Moved the ACE_HAS_REENTRANT_FUNCTIONS - into the section of #defines that should be commented out if - we're not building a multi-threaded library. - - * ace/Service_Config.cpp (open): Tightened up the comparison of - the key with the ACE_DEFAULT_LOGGER_KEY to use string equality - rather than pointer equality. Thanks to Stephen Coy - <stevec@magna.com.au> for reporting this. - -Wed Feb 25 10:50:17 1998 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_osf1_4.0.GNU: added suppression of - "no corresponding delete" warning, until that is fixed in libTAO. - Added -lm to LIBS for MT_Cubit client. - - * include/makeinclude/platform_chorus_ghs.GNU: we can link now. - - * ACE-INSTALL.html: added information on Linux v. 2.0.x - getsockname () bug. - -Tue Feb 24 15:07:50 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * tests/test.{mdp,mak}: Added Env_Value_Test and OrdMultiSet_Test - into the workspace. - - Removed all odbc libraries from all projects. We don't really - need them to run ACE. Thanks to gonzo@ing.puc.cl (Gonzalo - A. Diethelm) for pointing this out. - -Tue Feb 24 07:14:16 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Asynch_Acceptor.cpp (handle_accept): Update the - event_handler's proactor pointer once it is accepted. Thanks to - Jason Katz <Jason@email.rts-inc.com> for reporting this. - - * apps/Gateway/Gateway/Event.h (class Subscription): Moved the - htonl of total_consumers_ to *after* the loop that uses it! - Thanks to Jorn Jensen <jornj@funcom.com> for reporting this. - -Tue Feb 24 05:08:31 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Asynch_Acceptor.i: Filled in some missing keywords. - - * ace/Log_Record.h (ACE_Log_Record): Hacked it to work for CE. - Undefined the ostream version of print (). I'm planning to drop - iostream support on WinCE. (Then, I don't have to write my - own. ;-) - - * ace/ACE.{h,i} (timestamp, format_hexdump): Removed UNICODE_ONLY version. - Changed to use ASYS_TCHAR to differentiate this condition. - - * ace/OS.h (ACE_OS): Undefined shm_{open,unlink} if - ACE_HAS_MOSTLY_UNICODE_APIS. - - * ace/OS.cpp (ctime,ctime_r): Added CE implementation. - - * ace/streams.h: Added /**/ between #include and <..> in several - places. - - * ace/*: Changed ACE_HAS_UNICODE_ONLY to - ACE_HAS_MOSTLY_UNICODE_APIS. - -Mon Feb 23 19:41:53 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Asynch_Acceptor.i (bytes_to_read): Added an inline accessor - for the bytes_to_read_ field. Thanks to Jason Katz - <Jason@email.rts-inc.com> for suggesting this. - -Mon Feb 23 12:26:06 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.cpp (sched_params): removed ACE_ADAPT_RETVAL from PTHREADS - ::sched_setscheduler () call, because it returns the old scheduling - policy, not 0, on success. - - * ace/config-osf1-4.0.h: removed ACE_HAS_EXCEPTIONS because that is - enabled by platform_osf1_4.0{,_rcc}.GNU. - - * include/makeinclude/platform_osf1_4.0.GNU: added suppression of - warning 835 because /usr/include/cxx/memory, line 505, triggers - a warning about an unreferenced parameter. - - * include/makeinclude/platform_osf1_4.0{,_rcc}.GNU: added - "exceptions = 1" so that ACE exception handling can be - disabled on the make command line. - -Mon Feb 23 11:27:10 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE version 4.4.30, released Mon Feb 23 11:27:10 1998. - -Mon Feb 23 11:11:18 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Process.h: Define ACE_Process_Options::NO_EXEC as 0 on - Win32. It has no effect on NT anyway. - -Sun Feb 22 21:37:43 1998 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_vxworks5.x_g++.GNU: updated - comments to reflect that WIND_BASE and WIND_HOST_TYPE - environment variables must be set. Thanks to Nanbor for - reporting that. Also, set CROSS-COMPILE flag. - - * include/makeinclude/rules.lib.GNU: updated VOBJS so that it - will build the fancy new TAO/orbsvcs hierarchy. - - * include/makeinclude/platform_{chorus*,lynxos,vxworks5.x_g*}.GNU: - set CROSS-COMPILE flag. It's only used by TAO at this point. - - * ace/Svc_Conf_y.cpp: updated based on new Svc_Conf.y. - - * etc/Svc_Conf_y.cpp.diff: regenerated for latest Svc_Conf_y.cpp. - - * ACE-INSTALL.html,include/makeinclude/platform_lynxos.GNU: moved - LynxOS installation info from its platform file to ACE-INSTALL.html. - -Sun Feb 22 21:26:19 1998 James CE Johnson <jcej@chiroptera.tragus.org> - - * tests/IOStream_Test.cpp (server): break out of while loop if - read from client_handler fails. - -Sun Feb 22 13:06:20 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Asynch_Acceptor.h (class ACE_Asynch_Acceptor): Made the - accept method virtual so that users can override it and do - things like enforce limits on the maximum number of simultaneous - connections. Thanks to Jason Katz <Jason@email.rts-inc.com> for - suggesting this. - -Sun Feb 22 13:22:30 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.h: Moved definition of ACE_MAP_{PRIVATE,SHARED,FIXED} - down to where every platform can see. - -Sun Feb 22 13:06:20 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Svc_Conf.y: Fixed a couple of places where we were - potentially dereferencing NULL pointers when services failed to - load. Thanks to Eric C. Newton <ecn@smart.net> for reporting - this. - -Sat Feb 21 12:44:51 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE version 4.4.29, released Sat Feb 21 12:44:51 1998. - -Sat Feb 21 11:53:45 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Hash_Map_Manager.cpp (close_i): Remember to reset the - current size to 0 after deallocating all entries in the map. - Thanks to Alex for noticing this. - -Fri Feb 20 06:36:04 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Timer_Queue_Adapters (activate): added thread_names argument. - - * ace/OS.{h,i} (cond_timedwait): fixed on LynxOS (DCE_DRAFT4_THREADS), - where pthread_cond_timedwait () expects a relative instead of - absolute time. - - * tests/Buffer_Stream_Test (Consumer::svc): removed thr_yield for - LynxOS, now that ACE_OS::cond_timedwait () is fixed. - - * ace/OS.i (mmap): use ACE_DEFAULT_FILE_PERMS on LynxOS for the - shared memory object. - - * ace/Mem_Map.* (unmap): on LynxOS, write back the shared object to - the plain file, if the mmap () protection included PROT_WRITE. - - * ace/README: added ACE_HAS_BOOL, which can be used by TAO. - -Thu Feb 19 17:59:53 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * apps/JAWS/clients/Caching/*: Reworked and cleaned up these files a - bit. Added a test_URL.cpp file. Notice that these file will - finally be moved to other directory. - -Thu Feb 19 15:52:50 1998 Stanley D Leeson <STANLEY.D.LEESON@cdev.com> - - * ace/Task.{h,cpp} (activate): added optional thread_names array - argument. - -Thu Feb 19 12:55:00 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.h: Added /**/ between #include and '<', Thanks to - gonzo@ing.puc.cl (Gonzalo A. Diethelm) for providing the - patches. - -Thu Feb 19 11:18:08 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: added ACE_MAP_{PRIVATE,SHARED,FIXED}, and - ACE_OS::shm_{open,unlink}. - - * ace/OS.i: added ACE_OS::shm_{open,unlink}, and mmap support on - LynxOS. - - * ace/Mem_Map.*,Shared_Memory_MM.h,Filecache.cpp: default share - value to ACE_MAP_PRIVATE instead of MAP_PRIVATE, to allow easy - overriding in config files. - - * ace/config-lynxos.h: added ACE_HAS_SHM_OPEN, and map - ACE_MAP_PRIVATE to ACE_MAP_SHARED because MAP_PRIVATE is - not supported on LynxOS 2.5.0. - - * tests/Mem_Map_Test.cpp: replace file_size with line_length because - that's more accurate. - - * tests/SOCK_Connector_Test.cpp: use ACE_DEFAULT_SERVER_HOST on - LynxOS, and removed #include of sys/utsname.h because it's in OS.h. - - * include/makeinclude/platform_lynxos.GNU: added LYNXOS so that - other Makefiles can easily know if where building for LynxOS. - - * ace/ACE.cpp (recv_n): commented out the MSG_PEEK flag check because - it caused IOStream_Test to not terminate. - - * ace/Env_Value_T.{h,i}: moved ACE_Convert specializations from .i - file to .h file, because the resulted in multiple instantiations - with ACE_INLINE disabled, and missing specializations without - ACE_HAS_TEMPLATES_REQUIRE_SOURCE and inlining enabled. - - * tests/TSS_Test.cpp: weakened to work with LynxOS. It has a very - limited number of keys, maybe 16, and can't free them. And, the - test always core dumps if we try to request more keys than it has - available. - - * tests/run_tests.sh: disabled some tests on LynxOS. - -Thu Feb 19 10:41:05 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/SOCK.cpp (open): It appears that we don't need to do - SO_REUSEADDR if protocol_family != PF_UNIX. Thanks to Chuck - Gehr <gehr@sweng.stortek.com> for reporting this. - -Wed Feb 18 22:34:40 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * examples/Reactor/Dgram/Dgram.dsw: - * examples/Reactor/Dgram/Dgram.dsp: - * examples/Reactor/Dgram/CODgram.dsp: Newly added files for NT. - - * examples/Reactor/Dgram/Dgram.cpp: - * examples/Reactor/Dgram/CODgram.cpp: Made them runnable on NT. - - * ace/Process.h (ACE_Process_Options): Changed the definition of - ACE_Process_Options::NO_EXEC to 0 for Win32 because it doesn't - mean anything on NT (you must exec a new program on NT, no fork - syntax available.) With this, we can simulate the fork effect - more or less (still not portable thou. See - examples/Reactor/Dgram/ for some examples.) - -Wed Feb 18 16:07:14 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * netsvcs/lib/Log_Message_Receiver.cpp (log_output): There was a - problem where "outputfile" can be NULL, which causes problems - when it is dereferenced. Thanks to Stephen Coy - <stevec@magna.com.au> for reporting this. - - * ace/ACE.cpp (ldfind): modified the code to allow it to - consistently return a status indicating whether or not a - potential shared library file was found. Thanks to Patrick - J. McNerthney <pat@thememedia.com> for reporting this. - -Wed Feb 18 15:30:48 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * include/makeinclude/rules.lib.GNU: - Now we can create libraries with some files in - subdirectories. Though the Makefile still needs to set VPATH. - -Wed Feb 18 12:22:48 1998 David L. Levine <levine@cs.wustl.edu> - - * tests/Buffer_Stream_Test.cpp (Consumer::svc): added a - thr_yield so that the test magically works on LynxOS. - ACE_OS::cond_timedwait is busted on LynxOS, we're investigating it. - -Wed Feb 18 11:53:00 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * bin/auto_compile: - Added netsvcs to auto_compile list. - -Tue Feb 17 18:27:15 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.28, released Tue Feb 17 18:27:15 1998. - -Tue Feb 17 17:59:32 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * examples/Reactor/Multicast/Multicast.dsw: - * examples/Reactor/Multicast/client.dsp: - * examples/reactor/multicast/server.dsp: New project files. - -Tue Feb 17 15:53:41 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/ACE.cpp (recv_n): Added a check to drop out if the MSG_PEEK - flag is set! Thanks to Keith Brown <kalbrown@ix.netcom.com> for - reporting this. - -Tue Feb 17 12:56:24 1998 Chris Cleeland <cleeland@cs.wustl.edu> - - * include/makeinclude/wrapper_macros.GNU: Broke the "purify" - command out of PRELINK into its own variable: PURELINK. This was - done to quiet purify when building shared libraries. It wasn't - possible simply to remove the PRELINK reference from SOLINK.cc - because PRELINK is also used for specifying quantify. - - Also added two new variables: PURE_CACHE_DIR and - PURE_CACHE_BASE_DIR. PURE_CACHE_DIR is the value of the - purify/quantify '-cache-dir=' option, and PURE_CACHE_BASE_DIR is - used as the base directory to build a user/compiler-specific cache - directory. In general, you probably want to use them like this: - "make PURE_CACHE_BASE_DIR=~/mypurecache", which will end up - creating the directory ~/mypurecache-<username>-<compiler>. - -Tue Feb 17 09:23:05 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: there were two SIGCHLD #defines, if it was not - already #defined. I changed the second one to be SIGCLD instead. - - * ace/config-linux-common.h: removed ACE_LACKS_TCP_H on alpha because - it's not necessary with glibc-2.0.6. - -Tue Feb 17 08:41:06 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS.h (ACE_PROPER_SINGLETON_INSTANTIATION): Changed spelling - of SINGLETONE to SINGLETON. Thanks to Mark Maris - <maris@scgp.com> for reporting this. - -Tue Feb 17 07:41:32 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.27, released Tue Feb 17 07:41:32 1998. - -Tue Feb 17 03:32:30 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ACE.*: Changes for Windows CE just keeps going and going and - going... All following functions are member functions of class - ACE. - - This member funcion has wchar version _only_ when - ACE_HAS_UNICODE_ONLY is defined: - - timestamp - - I've added wchar version of the following member functions: - - strecpy - strsplit_r - strrepl - basename - execname - format_hexdump - - These functions are not defined on CE because they make no sense - on CE. - - strenvdup - daemonize - - There is no stdin defined on CE. Before we implement a stdin - for CE, these functions don't make any sense. - - read_adapter - register_stdin_handler - remove_stdin_handler - - Changing these functions to UNICODE will need to change the - svc.conf parser. Therefore, I'll leave them to a later point. - - ldopen - ldfind - - * ace/README: Added a new directive ACE_HAS_UNICODE_ONLY which - indicates that most APIs must use wchar instead of char. (Note - that on WinCE, only WinSock functions use the traditional char - interfaces. Most other functions only support wchar - interfaces, include str* functions.) - -Mon Feb 16 14:57:26 1998 David L. Levine <levine@cs.wustl.edu> - - * tests/Priority_Task_Test.cpp (open): fallback to minimum - ACE_SCHED_OTHER priority instead of 0. - - * tests/Priority_Task_Test.cpp (open): suppress warning message - about retrying task activation with fallback priority unless -d - option is specified. - - * ace/High_Res_Timer.cpp: initialize global_scale_factor_ to 1 if - ACE_HAS_{,AIX_}HI_RES_TIMER and ACE_WIN32 are not defined. - Thanks to Robert Olson <olson@mcs.anl.gov> for reporting a - problem with bad hires time values on Irix. - -Mon Feb 16 10:51:28 1998 Oliver Kellogg <Oliver.Kellogg@vs.dasa.de> - - * include/makeinclude/platform_osf1_4.0_rcc.GNU: added this - new platform file for Rational rcc (2.4.1) on DU 4.0. - - * ace/config-osf1-4.0.h: added ACE_LACKS_SIGNED_CHAR to - RATIONAL section. - - * ace/OS.h: added __rational__ to list of users of long - ACE_UNUSED_ARG. - - * ace/OS.i (gethostbyname_r): added some ACE_UNUSED_ARGs for - Digital Unix. - -Sun Feb 15 16:32:52 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * netsvcs/clients/Naming/Client/svc.conf: Removed "-s" flag, which - set the database name for local namespce. This was causing - grief if users forgot to set this environment variable. Thanks - to Johannes Gutleber <Johannes.Gutleber@cern.ch> for reporting - this. - -Sun Feb 15 15:55:47 1998 Darrell Brunsch <brunsch@cs.wustl.edu> - - * ace/Timer_Hash_T.{cpp,h}: Timer Hash wasn't keeping the earliest - node pointer up to date with cancels. - -Sat Feb 14 21:12:52 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Synch_T.h (class ACE_Guard): this->acquire() already sets - the owner, so we don't need to do it in the constructor. - -Sat Feb 14 20:57:59 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * netsvcs/lib/Log_Message_Receiver.{h,cpp} (log_output): Added this to - all receiver classes. It allows writing messages to a user - specified ostream. - - * netsvcs/lib/Server_Logging_Handler_T.cpp - (handle_logging_record): * netsvcs/lib/Client_Logging_Handler.cpp - (send): Use the above change to avoid missing data on NT. Thanks - to Greg White <gwhite@northlink.com> for providing the fix. - -Sat Feb 14 18:34:37 1998 Darrell Brunsch <brunsch@cs.wustl.edu> - - * ace/Timer_{Wheel,Hash}_T.cpp: Fixed a bug with earliest time - pointers not being set correctly. Thanks to Mark Evans - <mark.evans@tandem.com> for reporting this. - - * tests/Timer_Queue_Test.cpp: Added a quick test to check to - make sure the earliest time is reported correctly. - -Sat Feb 14 18:28:19 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/SOCK_Dgram_Mcast.cpp (subscribe): Fixed a problem with the - ACE_SOCK_Dgram_Mcast subscription method. Originally, each time - the ACE_SOCK_Dgram_Mcast subscribe() method was invoked, it - created a new socket, did appropriate setsockopt calls - (e.g. SO_USEPORT), did a bind() to that socket, and then did the - setsockopt call that actually joins the mcast group. The - problem with this was that is wasn't possible to a single socket - to subscribe to more than one multicast group. Multiple calls - to ACE_SOCK_Dgram_Mcast.subscribe() didn't work because the new - socket descriptor that was issued didn't listen to the old - multicast group(s). The fix was to change the subscribe() - method to only create, set up, and bind the socket if - get_handle() == ACE_INVALID_HANDLE. Thanks to Dave Meyer - <dmeyer@std.saic.com> for reporting this. - -Sat Feb 14 03:29:50 1998 Sergio Flores <sergio@tango.cs.wustl.edu> - - * ace/ACE.cpp (set_handle_limit): Fixed a couple of bugs - setting the maximum number of file descriptors. If we use - setrlimit, we should first get the hard limit before calling - it, or we get an EINVAL for trying to set a hard limit of 0 - and a soft limit greater than that. Also, for -1 the call - was not setting the limit to the maximum allowable. - - * ace/OS.h (ACE_MAX_DGRAM_SIZE): Moved the #define - ACE_MAX_DGRAM_SIZE from the PSOS specific part of OS.h. - -Fri Feb 13 19:25:55 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Log_Record.cpp (ACE_Log_Record): Changed the "set" method - for priority() so that it doesn't convert the priority into the - base two logarithm of the parameter. Instead, this is handled - by the "get" method of priority(). Thanks to Eric C. Newton - <ecn@smart.net> for reporting this. - - * ace/OS.h: Added a new #define called ACE_MAX_DGRAM_SIZE, which - is set to 8k by default. Naturally, this can be overwritten in - platform-specific config.h files. - -Fri Feb 13 10:40:36 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE version 4.4.26, released Fri Feb 13 10:40:36 1998. - -Thu Feb 12 14:33:12 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.i (mktemp): I missed this one in Win32 section. - -Fri Feb 13 09:04:13 1998 Oliver Kellogg <Oliver.Kellogg@vs.dasa.de> - - * ace/config-osf1-4.0.h: added Rational RCC support. - -Wed Feb 11 09:50:27 1998 Steve Huston <shuston@riverace.com> - - * ace/config-dgux-4.x-ghs.h - * include/makeinclude/platform_dgux4_ghs.GNU: Configs for ACE on - DG/UX 4 using Green Hills Multi C++ compiler, and draft 10 pthreads. - - * ace/README - * ace/OS.h: New macro ACE_NEEDS_SCHED_H for platforms which needs to - #include <sched.h> to get complete defs for thread scheduling. - - * ace/OS.cpp: Added a DG/UX-specific pthread_wait function for use with - draft 10 pthreads on DG/UX. - -Wed Feb 11 09:32:10 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE-INSTALL.html: updated DU CXX and Linux info. - - * ace/Makefile: moved INSTALL= to before include of rules.local.GNU, - so that the (unused) install command doesn't show up with make -n. - - * include/makeinclude/platform_chorus_ghs.GNU: put posix -I before - stdc -I. - - * ace/OS.h: removed include of stdfileio.h on ghs/Chorus. It's not - necessary with the above change to platform_chorus_ghs.GNU. - -Wed Feb 11 09:24:51 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE version 4.4.25, released Wed Feb 11 09:24:51 1998. - -Tue Feb 10 14:58:00 1998 Chris Gill <cdgill@cs.wustl.edu> - - * ace/Containers.cpp : removed unreachable break; - -Tue Feb 10 09:34:27 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/INET_Addr.cpp (get_host_name): moved declaration of local - "hentry" so that it's not seen on Chorus. - - * ace/OS.i (getprotoby{name,number}{,_r}): not supported on Chorus. - - * ace/Makefile: build Select_Reactor.o without -g on ghs/Chorus, - because it causes GreenHills 1.8.8 to core dump. - - * ace/config-chorus.h: adjusted to work with GreenHills 1.8.8. - - * include/makeinclude/platform_chorus_ghs.GNU: use Gnu ar instead - of ax, because we don't have a license for ax. - - * Timer_Queue_Adapters.{h,i} (activate): added stack and stack_size - parameters to match declaration in ACE_Task_Base. - - * include/platform_linux{,_lxpthread}.GNU: removed - -fno-strict-prototypes from CFLAGS because it is no longer needed. - - * include/makeinclude/platform_sunos5_g++.GNU: added -lposix4, to - be consistent with the Sun C++ platform file. Some of the TAO - tests need it. - -Mon Feb 09 18:10:37 1998 Chris Gill <cdgill@cs.wustl.edu> - - * ace/Containers.{cpp,h} : made next () a const member function in - ACE_Ordered_MultiSet_Iterator<T> - -Mon Feb 09 14:52:25 1998 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_chorus_ghs.GNU: added include path. - - * ace/config-chorus.h: with GreenHills, removed - ACE_HAS_WCHAR_TYPEDEFS_CHAR and ACE_LACKS_UNISTD_H. We'll - see if --multibyte_chars --wchar_t_keyword work. - - * ace/OS.h: added #include <stdfileio.h> with GreenHills/Chorus. - - * netsvcs/clients/Naming/Client/Client_Test.cpp (handle_input): - get input from stdin, not ACE_STDIN. Thanks to Jack Erickson - <JErickson@exchange.ml.com> for reporting this. - -Sun Feb 8 15:33:33 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * include/makeinclude/wrapper_macros.GNU: - * include/makeinclude/rules.lib.GNU: - * include/makeinclude/rules.local.GNU: Rename - ACE_LDSO_Major_Minor_Version_Number to ACE_LDSO_Version_Number. - - * include/makeinclude/rules.local.GNU: Put back checking of - current directory into library installation rules but with an - extra check for null ACE_LDSO_Version_Number. - -Sun Feb 08 12:06:30 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE-INSTALL.html: added Scott Snyder's egcs 1.0.1 patch, which - allows netsvcs/lib/Server_Logging_Handler.cpp to build on - Linux/RedHat 5.0. - - * include/makeinclude/wrapper_macros.GNU: protected ACE_ROOT... - additions to INCLDIRS and LDFLAGS so they're only added once. - Thanks to James CE Johnson <jcej@lads.com> for this addition. - - * ace/config-linux-common.h: added ACE_HAS_TERM_IOCTLS. Thanks - to James CE Johnson <jcej@lads.com> for this addition. - - * examples/Naming/test_multiple_contexts.cpp,test_non_existent.cpp, - test_writers.cpp: - replaced uses of cout with ACE_DEBUG. - - * examples/performance-tests/Misc/basic_perf.cpp, - context_switch_time.cpp,preempt.cpp: - divide ACE_hrtime_t by ACE_UINT32, to prevent ambiguity with - ACE_U_LongLong operator / overloading. - -Sun Feb 8 01:45:59 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * include/makeinclude/wrapper_macros.GNU (versioned_so): Added - a new directive to allow automatic installing of so libraries - with version number. This flag is default to 0. Put this line, - - versioned_so = 1 - - in your platform_*.GNU file to enable this feature. - - (ACE_LDSO_Major_Minor_Version_Number): Define the version number - to be appended to versioned so libraries (Notice: it include a - dot at the beginning.) If versioned_so == 0, this macro expand - to nothing. - - * include/makeinclude/rules.lib.GNU (INSTALL): - * include/makeinclude/rules.local.GNU (INSTALL rules): Modified - installation rules to allow installing so libraries with version - numbers. - - * include/makeinclude/platform_netbsd.GNU: - * include/makeinclude/platform_freebsd[_pthread].GNU: Enabled - new versioned_so feature. - -Sat Feb 07 05:51:15 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Synch_T.cpp (ACE_TSS): Changed to use MessageBox on WinCE. - Because there's no console in WinCE, we can't output warning - messages to stderr. - - * ace/README (ACE_HAS_UNICODE_ONLY): This new directive is used to - removed some char version function definitions because it - doesn't make too much sense to support these function when most, - if not all, strings in the system are wchar strings. - - * ace/OS.{h,i} (ACE_DEFAULT_LOGGER_KEY): Added wchar definition if - we are using UNICODE. - (ACE_SHE_{TRY,FINALLY,EXCEPT}): WinCE doesn't support Win32 - structual exception, so these macros should be defined as on - UNIX platforms. - (mktime): Removed this function from WinCE. Don't even think - about using them. - (ACE_HAS_UNICODE_ONLY): If this directive is defined, the char - version of the following functions are not defined in ACE_OS: - - chdir - mkdir - mkfifo - mktemp - getcwd - unlink - tempnam - - Some of them has wchar version. - - (chdir,mkdir,getcwd): Added wchar version. Unfortunately, getcwd - still doesn't work for WinCE. - - * ace/OS.cpp (mktemp): Added wchar version of this function. - Other changes added. Please refer to the entry above. If - platform/compiler supports this function, then both char and - wchar versions are defined as inline functions in OS.i. - (socket_{init,fini}): Removed leading :: from WSAStartup and - WSACleanup because these two are implemented as macros on - WinCE. Changed to use MessageBox instead of cerr for outputing - error messages. - - * ace/config-win32-common.h: Don't defined ACE_HAD_P_READ_WRITE on - WinCE. - - * ace/config-WinCE.h: Defined directive ACE_HAS_UNICODE_ONLY. - - * ace/ace_dll.dsp: Adjusted CE project settings. - -Fri Feb 06 15:28:11 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-sunos5.5-sunc++-4.x.h: disable explicit template - instantiations with CC 4.2, because it still does implicit - instantiation. We need to figure out how to disable that. - Then, we can enable -instances=global to get the real - benefits of explicit template instantiation with CC. - - * tests/Barrier_Test.cpp: updated a comment re. the VxWorks - task names test. - -Fri Feb 06 10:42:22 1998 David L. Levine <levine@cs.wustl.edu> - - * ACE version 4.4.24, released Fri Feb 06 10:42:22 1998. - -Fri Feb 06 10:30:00 1998 Chris Gill <cdgill@cs.wustl.edu> - - * ace/Containers.cpp: fixed error in - ACE_Doubly_Linked_List_Iterator<T>::first (void); - -Fri Feb 06 08:27:32 1998 Chris Gill <cdgill@cs.wustl.edu> - - * ace/Containers.{cpp,h} : added "first" methods to all container - iterators: each positions the iterator at the first element - in the container and returns 1, or if the container is empty - positions the iterator at a well defined "null" position for - that container and returns 0: - - template <class T> int - ACE_Unbounded_Set_Iterator<T>::first (void); - - template <class T> int - ACE_Unbounded_Queue_Iterator<T>::first (void); - - template <class T> int - ACE_Unbounded_Stack_Iterator<T>::first (void); - - template <class T> int - ACE_Bounded_Set_Iterator<T>::first (void); - - template <class T> int - ACE_Fixed_Set_Iterator<T>::first (void); - - template <class T> int - ACE_Doubly_Linked_List_Iterator<T>::first (void); - -Wed Feb 04 22:23:24 1998 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_linux*.GNU: commented out or - removed SUPPRESS_DASH_G, because it's not needed with hacked - gas on Linux/Alpha. - - * ACE-INSTALL.html: added some comments on Linux/RedHat 5.0, - for both Intel and Alpha. Thanks to James CE Johnson - <jjohnson@lads.com>, Torbjorn Lindgren <tl@funcom.com>, - and Oleg Krivosheev <kriol@fnal.gov> for this information. - - * ace/config-linux-common.h: replaced #define of ACE_HAS_EXCEPTIONS - with a compile time check for egcs, which supports exceptions. - Exceptions can now be enabled on the make command line. Also, - moved some comments on Linux system configuration up to - ACE-INSTALL.html. - - * ace/config-linux-common.h,config-g++-common.h: moved - ACE_HAS_EXCEPTIONS #define from config-linux-common.h to - config-g++-common.h. - - * include/makeinclude/wrapper_macros.GNU: disabled exception handling, - the default, on g++ >= 2.8.0. On both Linux/alpha and - Solaris 2.5.1/sparc, exception handling increased the size of - libACE.so by about 25 percent. Exception handling can easily - be enabled on the make command line or by editing your - include/makeinclude/platform_macros.GNU. - - * netsvcs/lib/TS_Server_Handler.cpp: instantiate - ACE_Svc_Handler<ACE_SOCK_STREAM, ACE_NULL_SYNCH> if - ACE_HAS_STREAM_PIPES, not if ACE_HAS_TLI. Thanks to Oliver - Kellogg <Oliver.Kellogg@vs.dasa.de> for reporting this problem. - - * Makefile: made ACE-INSTALL a dependency of release and - cleanrelease so that it is only updated when necessary. - - * ace/OS.{h,i},README (gets): wrapped with !ACE_LACKS_GETS. - - * ace/config-g++-common.h: added ACE_LACKS_GETS for egcs. It - supports ::gets (), but warns about it at link time on Linux. - The Solaris gets () man pages recommends using fgets () instead. - - * netsvcs/clients/Naming/Client/Client_Test.cpp: replaced use of - ACE_OS::gets () with ACE_OS::fgets (). - -Wed Feb 04 11:07:46 1998 Chris Gill <cdgill@cs.wustl.edu> - - * ace/Containers.cpp : fixed my "fix" to new_position pointer in - ACE_Ordered_MultiSet<T>::find_from () - now checks the pointer - before assigning something to it. - -Tue Feb 03 15:11:00 1998 Chris Gill <cdgill@cs.wustl.edu> - - * ace/Containers.cpp : fixed order of member initialization list - in ACE_DNode ctor, added update of new_position pointer in - ACE_Ordered_MultiSet<T>::find_from () - -Mon Feb 02 21:26:30 1998 David L. Levine <levine@cs.wustl.edu> - - * tests/Map_Manager_Test.cpp (test_map_manager): print out - item.int_id_ instead of local i so that item gets used (CXX - warned that it wasn't). - - * include/makeinclude/platform_osf1_4.0.GNU: added suppression of - warning 610, nonoverriding_function_decl. - - * ace/config-chorus.h,config-freebsd*.h,,config-hpux-10.x-g++.h, - config-irix{5.3,6.x}-g++.h,config-osf1-4.0.h,config-m88k.h, - config-netbsd.h,config-psosim-g++.h,config-sco-4.2-nothread.h, - config-sco-5.0.0-mit-pthread.h,config-sunos4-g++.h, - config-sunos5.4-g++.h,config-unixware-2.*.h: - migrated to config-g++-common.h. - - * ace/config-sunos5.5-egcs*.h: removed. config-sunos5.5-g++.h - now supports egcs, via config-g++-common.h. - - * ace/config-sco-5.0.0*.h: added RCS keyword string. - - * include/makeinclude/wrapper_macros.GNU: support "exceptions" - make flag on all platform. If enabled, it adds - -DACE_HAS_EXCEPTIONS to CCFLAGS. Not all platforms support - exception handling, of course, and that is noted in the - description of the "exceptions" make flag. - - * include/makeinclude/platform_sunos5_sunc++.GNU, - platform_sunos5.6_sparc_sunc++_4.2.GNU: - removed -DACE_HAS_EXCEPTIONS, because it's now done in - wrapper_macros.GNU. - - * examples/Reactor/Misc/test_event_handler_t.cpp: renamed - ACE_Sig_Handler to ACE_Test_Sig_Handler to avoid collision - with ACE_Sig_Handler in ACE library. - - * tests/Env_Value_Test.cpp (main): print out signed instead of - unsigned values. - -Mon Feb 02 14:40:00 1998 Chris Gill <cdgill@cs.wustl.edu> - - * tests/OrdMultiSet_Test.cpp: added explicit template instantiation - for ACE_DNode<int> - - * tests/OrdMultiSet_Test.dsp: new MSVC++ project for the test - - * tests/tests.dsw: added OrdMultiSet_Test.dsp project to workspace - - * tests/run_tests.{bat,sh,psosim,vxworks} : added OrdMultiSet_Test - - * tests/Makefile : added OrdMultiSet_Test - -Mon Feb 02 02:42:30 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.h (ACE_NOTSUP_RETURN): Moved its definition to the end of - this file where __TEXT has already been defined. WinCE needs to - use wchar. - - (struct utsname): Moved this definition after all datatypes are - filled in. Reverted my previous changed on it. - - Win32 only: Added wchar version of ACE_OS::{perror,fprintf, - vsprintf,sprintf}. - - Windows CE only: Undefined some char version function in class - ACE_OS. They are of no use on CE anyway. - - * ace/OS.i (fdopen): Added wchar version of ACE_OS::fdopen, it is - only available on Win32. - - Windows CE only: Added implementation for c-styled file - functions. Type FILE * is mapped to Win32 HANLDE under the hood - and Win32 API's are used to emulate these functions. - - * ace/OS.cpp: Removed some char version functions and added some - wchar functions. - -Sun Feb 01 15:12:49 1998 Chris Gill <cdgill@cs.wustl.edu> - - * tests/OrdMultiSet_Test.cpp: added an ACE test for the - ACE_Ordered_MultiSet and ACE_Ordered_MultiSet_Iterator - class templates. - - * ace/Containers.i (ACE_Ordered_MultiSet<T>::is_empty ()): - changed to return 1 rather than cause a SEGV when the - container is in fact empty (the above test discovered the bug). - -1998-02-01 Nanbor Wang <nanbor@fennec.wolfpack.cs.wustl.edu> - - * ace/OS.h (struct utsname): Separated WCE's definition from - ACE_LACKS_UTSNAME_T bacause VxWorks doesn't have TCHAR type. - -Sun Feb 01 12:09:11 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Containers.cpp (ACE_Ordered_MultiSet::insert and find): - expanded ITERATOR type of second argument, because g++ couldn't - deal with it. It can in the class declaration in Containers.h, - so I didn't change that. - - * ace/config-sunos5.5-g++.h,config-lynxos.h,config-vxworks5.x.h: - migrated to config-g++-common.h. - - * tests/run_tests.vxworks: disabled Message_Queue_Test because - it sometimes page faults or hangs the target. - -Fri Jan 30 15:35:30 1998 Chris Gill <cdgill@cs.wustl.edu> - - * ace/Containers.{cpp,h,i} : added class templates - ACE_Ordered_MultiSet and ACE_Ordered_MultiSet_Iterator - which respectively implement abstractions for containment - of and iteration over an ordered set that allows multiple - identical instances of a given item. - -Fri Jan 30 02:42:30 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.cpp (uname): Made it comform with Unicode. - - * ace/OS.h (struct utsname): Changed the internal string type to - TCHAR. - - * ace/OS.i (gettimeofday): Added Windows CE implementation. - -Thu Jan 29 21:02:24 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-sunos5.5-g++.h: revert so that ACE_HAS_STRING_CLASS - is #defined, regardless of g++ version. - - * ace/config-g++-common.h,config-gcc-2.7.2.h: replaced - config-gcc-2.7.2.h with config-g++-common.h, which will support - all g++ versions, including egcs. - - * ace/config-linux-common.h,config-sco-5.0.0-{fsu-pthread,nothread}.h: - migrated to config-g++-common.h. - -Thu Jan 29 20:28:30 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.23, released Thu Jan 29 20:28:30 1998. - -Thu Jan 29 16:17:28 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Log_Msg.cpp (open): Always close down the logger connection - before changing the setting. Added codes to allow resetting - logging endpoint back to STDERR from LOGGER. - - * ace/OS.h (ACE_DEFAULT_LOGGER_KEY): Changed to localhost:10012 - so it can be used in both netsvcs/lib/ and netsvcs/client/. - - * netsvcs/client/Logger/indirect_logger.cpp: Added code to reset - logging endpoint back to stderr. - - * netsvcs/lib/Client_Logging_Handler.cpp - (ACE_Client_Logging_Acceptor): Changed to initialize - logger_key_ with ACE_DEFAULT_LOGGER_KEY. - - * netsvcs/lib/Client_Logging_Handler.h: Removed DEFAULT_LOGGER_KEY - definition. - -Thu Jan 29 12:32:32 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * include/makeinclude/platform_irix6.x-sgic++.GNU: - Added support for debugging, somehow the DCCFLAGS were missing. - -Thu Jan 29 10:52:34 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-lynxos.h: added putenv () declaration. - - * include/makeinclude/platform_sunos5_g++.GNU: added comment - about disabling optimization with g++ 2.8.0. - - * ace/config-sunos5.5-g++.h: don't #define ACE_HAS_STRING_CLASS - with g++ 2.8.0. - - * include/makeinclude/platform_sunos5_egcs.GNU: include - the g++ platform file, it works just fine. The net effect - on egcs is these three changes: - - 1) -fno-implicit-templates is included in CPPFLAGS. That - option is necessary with the egcs (egcs-2.90.23 980102 - (egcs-1.0.1 release)) that we use to save _much_ text - space. - - 2) Use -G instead of -shared in SOFLAGS. That way, - libstc++.so, etc., need not be built. - - 3) The compiler is called g++ instead of c++. I don't - think this hurts anything. And Purify knows about g++, - but not g++. - - These (and any other) options can be overridden in - platform_sunos5_egcs.GNU if necessary. See the commented - example there. - - * ace/config-sunos5.5-egcs.h: added - ACE_HAS_EXPLICIT_TEMPLATE_INSTANTIATION to match - -fno-implicit-templates in the platform_sunos5_egcs.GNU. - - * tests/Env_Value_Test.cpp: updated comment to explain why we - can't use spawn with ACE_HAS_NONSTATIC_OBJECT_MANAGER. We - can, in general, but we can't on VxWorks because it doesn't - have fork (). - -Thu Jan 29 10:32:28 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * include/makeinclude/platform_sunos5_sunc++.GNU: - The posix4 library is needed in some TAO examples. - - * ace/Module.h: - * ace/Message_Queue.h: - Workaround HP/C++ compiler bug, it does not support enums inside - templates so we move them into base classes. Thanks to Sundar - T.J <tjsundar@cisco.com> for reporting this problem. - -Thu Jan 29 10:14:03 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/FIFO.h: Revised the comment at the beginning of ACE_FIFO to - clearly indicate that it doesn't work portably to Win32. Thanks - to Thomas Hampson <thomas.hampson@lmco.com> for suggesting this. - -Wed Jan 28 15:36:55 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * bin/auto_compile: - We can use the top level Makefile on TAO now. - -Wed Jan 28 11:59:41 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.i (isatty): Added conversion from a Win32 HANDLE to a C - file handle. It works correctly now. Thanks to Chuck R. Gehr - <GehrCR@louisville.stortek.com> for noticing the problem. - -Tue Jan 27 20:32:07 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Message_Block: Added an accessor to retrieve the reference - count on an ACE_Data_Block. Thanks to Dave Meyer - <dmeyer@std.saic.com> for contributing this. - -Tue Jan 27 15:09:08 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/OS.i, - * ace/Timer_Queue_Adapters.cpp: - When using ACE_CLEANUP_???_POP macros, it was being called with - argument 1, which should be 0 (don't execute the cleanup). - Thanks to Arturo Montes <mitosys@colomsat.net.co> for reporting - argument 1, which should be 0 (don't execute the cleanup).this. - -Tue Jan 27 14:56:49 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Basic_Types.*: more cleanup: got rid of ACE_USES_ACE_U_LONGLONG, - replaced VXWORKS && ghs with ACE_HAS_WCHAR_TYPEDEFS_CHAR, and - factored out some common #defines. Rearranged ACE_SIZEOF_LONG_LONG - and ACE_UINT64 logic so that it works on NT. - -Tue Jan 27 08:28:34 1998 Steve Huston <shuston@riverace.com> - - * ace/Basic_Types.h: Simplified and changed so that the config.h file - can pre-specify the ACE_SIZEOF_* macros in case the preprocessor - has trouble calculating them. - -Tue Jan 27 06:45:07 1998 David L. Levine <levine@cs.wustl.edu> - - * examples/IOStream/server/iostream_server.cpp: for egcs, use - ACE_SOCK_IOSTREAM instead of its expansion in the ACE_Svc_Handler - template instantiation. - - * ace/Event_Handler_T.h,examples/Reactor/Misc/test_event_handler_t.cpp: - removed special-case support for ACE_HAS_SIGNFO_T, because ace/OS.h - declares a siginfo_t without it. - - * tests/run_tests.vxworks: enabled Priority_Reactor_Test and - Message_Queue_Test, because they appear to run successfully now. - Most of the time, anyways. Message_Queue_Test occasionally - pages faults; there appears to be a memory management problem. - -Tue Jan 27 02:38:16 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.{i,cpp}: - * ace/config-WinCE.h: More Windows CE insanities. - -Mon Jan 26 13:20:58 1998 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_osf1_4.0_g++.GNU: added $(ACELIB) - to SOFLAGS. Thanks to Oliver Kellogg <Oliver.Kellogg@vs.dasa.de> - for reporting the problem with building shared libraries on - g++/DU 4.0. - - * ace/config-linux-common.h: moved ACE_LACKS_TCP_H so that it is - only seen on alphas (with glibc2). Thanks to Edgar Villanueva - <edgarvil@ix.netcom.com> for reporting this. - - * performance-tests/Misc/childbirth_time.cpp (prof_native_thread, - prof_ace_os_thread): changed type of local j to int to avoid - signed/unsigned comparison. - - * tests/Reactors_Test (Test_Task): changed type of member - handled_ from int to size_t to avoid signed/unsigned comparison - in Test_Task::handle_input (). - -Mon Jan 26 11:35:54 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * bin/auto_compile: - Added TAO_ROOT to the environment. - -Mon Jan 26 10:28:57 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.22, released Mon Jan 26 10:28:57 1998. - -Mon Jan 26 02:53:10 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.h: Namespace related macros are now ifdef'ed by - ACE_HAS_BROKEN_NAMESPACES instead of ACE_ORBIX_HAS_NAMESPACES. - - * ace/README - * ace/config-win32-common.h: Added a new flag - ACE_HAS_BROKEN_NAMESPACES to deal with MSVC's broken namespace - support. - - * ace/config-win32.h: Re-enable defining ACE_ORBIX_HAS_NAMESPACES - for Win32. - - * ace/config-WinCE.h: Temporary file to hold WinCE specific - configurations. It will probably be merged with - config-win32-common.h in some point. - - * ace/ace_dll.dsp: Added WinCE x86 emulation mode project - definition. - - * ace/OS.{h.i} (ACE_OS): Added Windows CE macros madness. - - * ace/config-win32-common.h: Ifdef'ed out ACE_HAS_SIG_ATOMIC_T for - WinCE. Prevent repeat definition of STRICT. - -Sun Jan 25 17:31:49 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/config-win32.h: Disabled setting ACE_ORBIX_HAS_NAMESPACES by - default on NT. MSVC's namespace support is buggy. - -1998-01-25 Nanbor Wang <nanbor@fennec.wolfpack.cs.wustl.edu> - - * tests/Conn_Test.cpp: Rearranged explicit template instantiations - order to avoid errors from egcs. Thanks to Gonzo for reporting - the bug. - -Fri Jan 23 12:24:05 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS.i (sleep): Fixed a typo where tv_sec was being set rather - than tv_usec. Thanks to Gonzo for pointing this out. - -Fri Jan 23 06:27:51 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/IOStream.h: use std::string if - ACE_USES_STD_NAMESPACE_FORSTDCPP_LIB IOStream.h. Thanks - to Gonzo <gonzo@ing.puc.cl> for the patch. Also, changed - DEC_CXX to __DECCXX_VER because that's more accurate. - - * tests/Barrier_Test.cpp: specify thread name and stack size - on VxWorks. Thanks to Stan Leeson <STANLEY.D.LEESON@cdev.com> - for suggesting this. - - * tests/Env_Value_Test.cpp (TEST_THIS): cast val to (type) - before casting to (u_int), to try to avoid DEC CXX compiler - warning about integer overflow. Thanks to Frederic Andres - <andres@rd.nacsis.ac.jp> for reporting this. - - * ace/config-osf1-4.0.h: added ACE_HAS_EXCEPTIONS, etc., with - CXX 6.0. Thanks to Torbjorn Lindgren <tl@funcom.com> for - the patch. - - * ace/OS.i (thr_kill): removed ACE_ADAPT_RETVAL because - ::taskNameToID returns the task ID, not success/fail. - Thanks to Richard Keizer <RICHARD.L.KEIZER@cdev.com> for - diagnosing the problem and providing the fix. - - * tests/run_tests.vxworks: enabled Thread_Manager_Test and - Message_Block_Test now that they succeed. Thanks to Richard Keizer - <RICHARD.L.KEIZER@cdev.com> for the ACE_OS::thr_kill fix. - - * ace/OS.cpp (thr_create), ace/Thread_Manager.cpp (kill_thr): - on VxWorks, use strncmp instead of strcmp to check for allocated - thread ID. - -Thu Jan 22 16:34:18 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * examples/Connection/misc/Connection_Handler.cpp (svc): Fixed a - bug in the example where the global Singleton Reactor was being - used rather than the object's Reactor. Thanks to Peter Gross - <pag@kokako.probita.com> for reporting this. - -Thu Jan 22 09:04:13 1998 David L. Levine <@cs.wustl.edu> - - * ace/Hash_Map_Manager.cpp (ACE_Hash_Map_Iterator_Base ctor): - added a cast of mm.total_size_ to ssize_t, to try to avoid - a compiler warning. Thanks to Brian Mendel <bmendel@mdc.com> - for reporting this. - - * ace/config-osf1-4.0.h: updated g++ support. Thanks to - Oliver Kellogg <Oliver.Kellogg@vs.dasa.de> for the patch. - - Thanks to Tom Venturella <tventurell@mdc.com> and Dave Mayerhoefer - <davem@lynx.com> (!) for requesting the following three changes: - - * ace/Thread_Manager.cpp: added support for a user-supplied - thread (VxWorks task) name. By itself, that means just passing - the ACE_thread_t along to ACE_OS::thr_create (). The tricky - part was then handling the case where the user doesn't supply - a name. It would be cleaner if the ACE_thread_t and ACE_hthread_t - typedefs were swapped on VxWorks, because the ACE_Thread_Manager - keys off of the ACE_thread_t. And in addition to being cleaner, - it would be safer because ACE_thread_t's currently aren't - guaranteed to be unique on VxWorks. - - * ace/OS.cpp (thr_create): on VxWorks only, use the t_id (task - name) and stack_size if supplied. Also added support for setting - the stack, but commented it out because it doesn't seem to work. - - * tests/Conn_Test.cpp: on VxWorks, assign thread (VxWorks task) - names to server and client threads, to test that feature. - - * ace/OS.cpp (spa [VxWorks only]): use ::taskSpawn () instead - of ::sp () so that the task name, ace_main, can be specified. - - * ACE-INSTALL.html: added note about not linked the ACE library - into every VxWorks executable. It should work when running - multiple ACE executables on a target, though I haven't tried it. - - * include/makeinclude/rules.local.GNU: set SOEXT if it was not set, - so that it's always non-null for make depend. - - * ace/OS.h: #define ACE_THR_PRI_FIFO_DEF on all threaded platforms, - if not already defined in OS.h or in the individual platform config - file. It should be used by applications for a default real-time - thread priority. - - * ace/config-chorus.GNU: added GreenHills support. - - * include/makeinclude/platform_chorus_ghs.GNU: added platform file. - -Wed Jan 21 21:37:03 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Hash_Map_Manager.cpp (close_i): Removed the HPUX_11 - destructor madness by saving the location to be destructed in a - temporary reference instead of calling the destructor from the - array directly. Thanks to Istvan Buki <ibuki@fedex.com> for - providing the patch. - -Wed Jan 21 20:36:43 1998 Chris Cleeland <cleeland@cs.wustl.edu> - - * ACE version 4.4.21, released Wed Jan 21 20:36:43 1998. - -Wed Jan 21 10:39:12 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Task.cpp (activate): for builds without threads only, - added ACE_UNUSED_ARG of stack and stack_size. - - * ace/High_Res_Timer.cpp: used unsigned constants instead of - signed, to allow resolution of overloaded - ACE_U_LongLong::operator/ (). - -Tue Jan 20 21:15:01 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.20, released Tue Jan 20 21:15:01 1998. - -Tue Jan 20 20:31:27 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * tests/test_config.h: - Changed several buffer sizes from BUFSIZ to MAXPATHLEN, they - were used for storing pathnames, so the later is a proper size - for them. - -Tue Jan 20 15:54:44 1998 Darrell Brunsch <brunsch@cs.wustl.edu> - - * ace/Timer_Heap_T.cpp: Fixed a bug that was introduced on - Sunday because of the non-use of position 0 (we have to - grow the heap one node early). - - * ace/Timer_Queue_T.{h,cpp}: Added a return_node () method - which is used to return a ACE_Timer_Node back into the - control of the Timer Queue after it leaves via - remove_first (). - -Tue Jan 20 15:52:31 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.19, released Tue Jan 20 15:52:31 1998. - -Tue Jan 20 12:49:05 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Hash_Map_Manager.cpp (ACE_Hash_Map_Manager): Member table_ - needs to be initialized to prevent deleting invalid memories. - -Tue Jan 20 11:57:24 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * bin/auto_compile: - Added several TAO components to the compile list. - The regular expression for make errors was missing several of - them. - -Tue Jan 20 03:36:54 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Hash_Map_Manager.cpp (open): We should call close_i () to - ensure previously allocated memory gets release before we - allocate new one. - -Mon Jan 19 22:21:34 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * tests/Env_Value_Test.cpp (main): Added test codes for unsigned - short. - - * ace/Env_Value_T.i (ACE_Convert): Added conversion function for - unsigned short. - -Mon Jan 19 15:13:47 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * bin/Makefile: Fixed a problem with the CCFLAGS definition that - was causing circular macro expansion. Thanks to Philippe - O'Reilly <philippe@robot.ireq.ca> for reporting this. - -Sun Jan 18 14:32:42 1998 Darrell Brunsch <brunsch@cs.wustl.edu> - - * ace/Timer_Heap_T.cpp: Problems were arising because of the - use of negative numbers in the timer_id array of the timer - heap for the free list. When the 0th index is in the free - list, pointers to it will show up as 0 (because 0 == -0). - This will look the same as a timer_id for the top of the - heap, and if cancel () was attempted on this invalid - timer_id, bad things would happen. To prevent this, index - 0 is now ignored so it doesn't show up in the free list. - -Sun Jan 18 18:12:52 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * examples/Service_Configurator/Misc/*.ds?: Added new project - files for this example. - -Sun Jan 18 17:27:52 1998 Darrell Brunsch <brunsch@cs.wustl.edu> - - * ace/Timer_Wheel_T.cpp: Fixed a problem with Timer Wheel's - expire(). The problem occured when the handle_timeout() method - of the event handler used would return a 0. This causes the - Timer Queue to cancel all timers using that event handler. - Because Timer Wheel's expire caches the position of some timers - that will be expired, it was trying to expire timers that were - already cancelled. Now it checks the timer before it expires. - -Sun Jan 18 16:08:36 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Thread_Manager: Modified the signature to the spawn_n() - method so that it's possible to pass in an array of stack - pointers and stack sizes. Thanks to Thomas Venturella - <tventurell@mdc.com> for requesting this. - - * ace/Task: Modified the signature to the activate() method so - that it's possible to pass in an array of stack pointers and - stack sizes. Thanks to Thomas Venturella <tventurell@mdc.com> - for requesting this. - -Sun Jan 18 15:54:16 1998 Darrell Brunsch <brunsch@cs.wustl.edu> - - * ace/Timer_Wheel_T.h: - ace/Timer_Hash_T.h: - ace/Timer_Queue_T.h: - ace/Timer_Wheel_T.cpp: - ace/Timer_Hash_T.cpp: Added pure virtual get_first() method to - Timer_Queue_T. Added implementations to Timer_Wheel and Timer_Hash. - Timer_Heap and Timer_List already had implementations. - -Sat Jan 17 23:40:02 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * netsvcs/VCAUTO.MAK: - * netsvcs/lib/VCAUTO.MAK: Added directory information for easier - debugging. - -Sat Jan 17 16:10:04 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * examples/Service_Configurator/Misc/main.cpp (main): Replaced the - buggy use of ACE_Service_Config daemon; with - ACE_Service_Config::open(). We'll have to fix this properly - some day, but for now this is the best solution. Thanks to - Pawel Roman <proman@npac.syr.edu> for reporting this. - -Sat Jan 17 00:42:13 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/SString.cpp (substring): Fixed bug. count should be - initialized to length and not len_. - -Fri Jan 16 15:52:34 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/IOStream.h (ACE_Streambuf): - * ace/config-win32-common.h: - * ace/iosfwd.h: - * ace/streams.h: Applied IOStream patches from James CE Johnson - <jcej@chiroptera.tragus.org>. - -Fri Jan 16 01:42:52 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/OS.h: Changed the definition of ACE_CORBA_* macros so that - on non-Win32 platforms, they will be expand to the nested - classes. Thanks to Jonathan Reis <reis@minniemouse.cemax.com> - for reporting this. - - * ace/SString.h: Added the correct export macros. - -Thu Jan 15 15:45:48 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * etc/ACE-guidelines.html: Reformatted this page a bit. - - * ace/Message_Queue.cpp (open): The fix from Slawomir Kuzniar was - incorrect. I reverted the change. - -Thu Jan 15 12:43:23 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * /: Added a bunch of makefiles (*.MAK) for one button - make-and-test on NT. - - * Makefile (FILTER): Added filter to remove VC one button related - makefile from ACE release. - - * ace/ACE_Library.{mdp,mak}: Added Shared_Memory.cpp file into the - project makefile. - -Thu Jan 15 09:12:26 1998 David L. Levine <levine@cs.wustl.edu> - - * tests/Barrier_Test.cpp,Message_Block_Test.cpp,Priority_Task_Test.cpp, - Reactor_Exceptions_Test.cpp,Reactors_Test.cpp,Thread_Mutex_Test.cpp: - (main): changed some local variables from int to size_t to avoid - signed/unsigned comparison. - - * ace/OS.i,config-linux-common.h,README: aded ACE_HAS_SOCKLEN_T - support for Linux/glibc2, which has a socklen_t type. - - * include/makeinclude/rules.local.GNU (depend.local): produce all 4 - combinations of .obj, .shobj, .o, and .so dependency targets. This - should take care of the problem of not rebuilding when a dependency - has changed, after we regenerate all the Makefiles. - - * include/makeinclude/rules.local.GNU (depend.local): removed the sed - rule that wiped out each Makefile line ending with :. I don't know - why that rule was there. With it, if a Makefile had a target such - as clean:, it got deleted. Not any more. - - * many Makefiles: updated dependences with revised depend.local. - - * bin/g++dep: added lib/g++-include path to gcc -MM invocation, so - that gcc can find all of its headers. (It didn't help to replace - gcc with g++, though I would have expected that to solve the - problem.) - - * ace/Timer_Heap_T.cpp (get_first): fixed ACE_TRACE id string. - - * apps/Gateway/Gateway/Concrete_Connection_Handlers.cpp, - * apps/Gateway/Peer/Peer.cpp (recv): - changed type of HEADER_SIZE to size_t to avoid signed/unsigned - comparison. - -Wed Jan 14 20:41:35 1998 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/Signal.i (is_member): - Fixed typo in new cast, the type is sigset_t not sigset. - -Wed Jan 14 20:27:44 1998 <coryan@MILONGA> - - * ace/OS.h: - The new location for AF_ANY and AF_FILE was in the middle of a - big #if !ACE_WIN32, hence it was not working for NT, I moved it - further down and now NT is happy. - -Wed Jan 14 19:01:59 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Message_Queue.cpp (open): In Message_Queue::open - deactivated_ flag is being set to 0 instead of - WAS_INACTIVE. This is causing incorrect return value from - first-time call to activate(). Thanks to Slawomir Kuzniar - <kuzniar@Bear.COM> for reporting this. - - * ace/Signal.i (is_member): Added an ACE_const_cast(). Thanks to - Oleg Krivosheev <kriol@fnal.gov> for reporting this. - -Wed Jan 14 06:45:52 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.i: Fixed typo in ACE_Time_Value::operator const timeval * - definition. (select's): cast the ACE_Time_Value to a - const timeval * and then to a timeval * so that they will - compile. - - * tests/Time_Value_Test.cpp (test_ace_u_longlong): made divisors of - operator/ tests explicitly unsigned, to avoid overloading problems - now that there are two division operators. - -Tue Jan 13 16:31:48 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/OS.i: Changed the signature of ACE_Time_Value::operator - timeval *() to return operator const timeval *(). Thanks to - Oleg Krivosheev <kriol@fnal.gov> for reporting this. Changed - several other unsafe casts to use the ACE_const_cast() macro. - -Tue Jan 13 13:30:23 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/Profile_Timer.cpp (elapsed_time): use floating point - arithmetic for real_time calculation, in case delta_t is - less than ACE_ONE_SECOND_IN_NSECS (and for more precision - otherwise). Thanks to Brian Mendel for tripping over this. - - * ace/Basic_Types.{h,i} (ACE_U_LongLong): added - operator/ (const double). - - * ace/OS.h: moved definition of AF_FILE, etc., to _after_ the - #include of sys/socket.h. That way, the #ifndef AF_FILE - protection will work properly on Linux/glib2. Thanks to - Amos Shapira <amos@gezernet.co.il> for breaking silence and - reporting this. - -Mon Jan 12 20:19:54 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/SString.i: Replaced all uses of NULL with 0 since NULL - causes problems with some lame C++ compilers. - - * ace/SString.i: Moved one of the inline find() methods down - so that GCC doesn't complain anymore. - - * tests/Reactor_Timer_Test.cpp (main): Changed long i to size_t i - to be consistent. - - * ace/Select_Reactor.h: Updated the explanation of how the - ACE_Select_Reactor_Token works. Thanks to Detlef Becker for - explaining this in great detail. - - * tests/test_config.h: Changed the type of all the consts to - size_t rather than int. This may cause a bit of upheaval at - first since we'll need to fix some other types. But it's the - "Right Thing"[TM] to do. - -Mon Jan 12 18:31:23 1998 Nanbor Wang <nanbor@cs.wustl.edu> - - * Start renaming MSVC 4.2 workspace/makefile names. - - * apps/Gateway/Peer/Peer4.{mdp,mak}: - * apps/Gateway/Gateway/Gateway4.{mdp,mak}: Made them in synch with - VC 5.0's status. - -Mon Jan 12 12:30:34 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/config-linux-common.h: moved - ACE_LACKS_STATIC_DATA_MEMBER_TEMPLATES so that it isn't used - with egcs. Thanks to Gonzo for testing this. - - * tests/Service_Config.test: changed type of VARIETIES to u_int to - avoid signed/unsigned comparison in main. Thanks to Gonzo for - reporting this. - - * ace/config-linux-common.h: added ACE_HAS_TEMPLATE_TYPEDEFS and - ACE_HAS_TYPENAME_KEYWORD. Thanks to Oleg Krivosheev - <kriol@fnal.gov> for reporting these. - - * ace/config-vxworks5.x.h: merged the two VxWorks configs (g++ and - ghs) into this new config for easier maintenance. Added - ACE_THR_PRI_FIFO_DEF, set to 6. - - * tests/Reactor_Timer_Test.cpp (main): cast i to u_long to avoid - signed/unsigned comparison. - -Sun Jan 11 23:02:07 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * tests/SString_Test.cpp (main): Removed the "silly" strings from - the tests. They are annoying and cause memory leaks. - -Sun Jan 11 22:16:28 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * apps/Gateway/Peer/Peer.{h,cpp} (transmit): Changed the function - to return the result of transmission. - (transmit_stdin): Make sure the function returns proper results - of transmission. - -Sun Jan 11 16:18:15 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.18, released Sun Jan 11 16:18:15 1998. - -Sun Jan 11 00:00:26 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * apps/Gateway/Gateway/Gateway.cpp (parse_connection_config_file): - Fixed a typo, == should be =. Thanks to David Levine for - reporting this. - - * apps/Gateway/Peer/Peer.cpp (recv), - apps/Gateway/Gateway/Concrete_Connection_Handlers.cpp (recv): - Added a cast to remove a GCC warning about signed/unsigned - comparisons. Thanks to David Levine for reporting this. - - * ace/Log_Msg.cpp (log): Made a minor change to move where the - message is copied from thread-specific storage into the transfer - buffer. Thanks to Patrick J. McNerthney <pat@thememedia.com> - for reporting this. - - * apps/Gateway/Peer/Peer: Added new logic to transmit subscription - requests to the Gateway. - - * apps/Gateway/Gateway/Event.h (class Event_Header): Make sure - that we correctly encode/decode the connection_id. - -Sat Jan 10 18:17:06 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * apps/Gateway/Gateway/Event.h: Added the event priority_ field to - the Event_Key. We don't currently use this, but we may some - day. - - * apps/Gateway/Gateway/Concrete_Connection_Handlers.cpp (put): - Changed the forward() method to process() since our new scheme - will need to do more than just forward, e.g., it will need to - handle subscriptions. - - * apps/Gateway/Gateway/Connection_Handler: Made the accessors - const member functions and also added accessors for the - Event_Channel * so this can be set by the - Connection_Handler_Acceptor::accept_svc_handler() method. - - * apps/Gateway/Gateway/Connection_Handler: Added mutators so that - we can set the local and remote addresses outside of the - constructor. - - * apps/Gateway/Gateway/Options: Added a connection_id() accessor - to set/get the connection id. This makes it possible to allow - Peers to connection with the Gateway. - - * apps/Gateway/Gateway/Gateway.cpp (parse_connection_config_file): - Added a sanity check to ensure that the connection ids start at - 1 and increment by 1. - -Sat Jan 10 22:37:37 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/SString.i (find): changed type of local i to size_t to - avoid signed/unsigned comparison. - -Sat Jan 10 14:57:01 1998 <coryan@MILONGA> - - * ace/Env_Value_T.h: - The return value for the assignment operator had no template - parameter, ditto for the arguments of the copy constructor and - the already mentioned operator. - -Sat Jan 10 13:20:08 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * tests/Conn_Test.cpp: Moved template instantiation codes around - to avoid warnings. - -Sat Jan 10 01:36:08 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/SString: Made the following changes to make the ACE string - class look more like the STL string class: - - - Added npos to all classes. - - - Added non-const reference operator[]. - - - Changed operator= and operator+= to return *this. - - - Added STL like find and rfind routines - - - Made the three classes be a similar as possible. - - - Changed the implementation of strstr to use the C library - strstr function. - - - Added constructors from single characters. - - - Made substring a const method. - -Fri Jan 9 19:11:39 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * tests/Map_Manager_Test.cpp (test_hash_map_manager): Added - ACE_UNUSED_ARG to eliminate warnings. - - * ace/Hash_Map_Manager.cpp (rbegin): Added more static cast to - eliminate warning messages. - -Fri Jan 9 11:35:53 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * examples/Service_Configurator/IPC-tests/server/server_test.cpp: - Fixed a syntax error. Thanks to John Mulhern - <9107@MN3.LAWSON.lawson.com> for reporting this. - - * ace/Log_Msg.cpp (log): Changed abort_prog == 0 to abort_prog so - that STDERR would be printed correctly. - - * ACE-categories: Added all the ACE files into the ACE-categories - listing. Thanks to Andres Kruse for motivating this. - - * include/makeinclude/platform_sunos5_egcs.GNU (CXX): Removed - -pipe from CC and CXX. It was already in CCFLAGS. - -Fri Jan 09 11:56:44 1998 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/config-win32.h (ACE_ORBIX_HAS_NAMESPACES): By default, you - will get the proper namespace usage for Orbix. If you don't - like this, comment out the #define ACE_ORBIX_HAS_NAMESPACES line - or #undef ACE_ORBIX_HAS_NAMESPACES in your config.h file after - including this config-win32.h. - -Fri Jan 09 07:04:47 1998 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_sunos5_g++.GNU: removed - -pipe from CC and CXX. It was already in CCFLAGS. - - * include/makeinclude/platform_vxworks5.x_g++.GNU: added - target-dependent support for symbol table extractor, and - disabled symbol table extraction (POSTLINK) by default. - Thanks to Keith Nicewarner <knicewar@sanders.com> for - reporting problems with this on PPC604 targets. - - * ace/config-linux-common.h: added ACE_HAS_STDCPP_STL_INCLUDES - and removed ACE_LACKS_ACE_IOSTREAM. Thanks to James CE Johnson - <jcej@lads.com> for reporting these. - - * tests/run_tests.vxworks: added Env_Value_Test. - -Fri Jan 09 00:14:20 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * apps/Gateway/Gateway/Connection_Handler_Acceptor.cpp: - * apps/Gateway/Gateway/Config_Files.cpp: Defined - ACE_BUILD_SVC_DLL flag. All declarations of an exported class - must be declared with the same exporting flag. - - * apps/Gateway/Peer/Options.h: Updated the path of include file - Event.h. - -Thu Jan 08 22:11:57 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * ace/Env_Value_T.i (ACE_Convert): Added static cast for int and - short conversion funcions. - -Thu Jan 08 21:37:30 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.17, released Thu Jan 08 21:37:30 1998. - -Thu Jan 8 19:50:43 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Task: Added a virtual destructor for ~ACE_Task_Base. Thanks - to Oleg Krivosheev <kriol@fnal.gov> for suggesting this. - - * ace: Added a new config-sunos5.5-egcs.h for the EGCS compiler. - Thanks to Oleg Krivosheev <kriol@fnal.gov> for contributing - this. - - * include/makeinclude: Added a new platform macros file for the - EGCS compiler on Solaris 2.x. Thanks to Oleg Krivosheev - <kriol@fnal.gov> for contributing this. - - * include/makeinclude/platform_sunos5_g++.GNU (CXX): Added the - -pipe flag. - - * man: Updated all the manual pages. - - * ace/Log_Msg: Added changes to allow the logging of custom built - log records. The main reason for these changes is to allow code - to handle the logging of out of memory conditions without - allocating any additional memory. Thanks to Patrick - J. McNerthney <pat@thememedia.com> for this feature. - - * ace/Log_Msg: Added a new method called log_priority_enabled(), - which returns true if the priority is enabled (gee, is *that* - what it does? ;-)). Thanks to Patrick J. McNerthney - <pat@thememedia.com> for this feature. - - * examples/Service_Configurator/IPC-tests/server/server_test.cpp - (main): Revised the driver program so that it doesn't create an - explicit ACE_Service_Config object. Thanks to John Mulhern - <9107@MN3.LAWSON.lawson.com> for reporting this. - - * apps/Gateway/{Gateway,Peer}/Options.cpp: Oops, ACE_BUILD_DLL - should be ACE_BUILD_SVC_DLL, yikes! Thanks to Jeff Hellzen - <JHellzen@hwdcsaws.cahwnet.gov> for reporting this. - - * apps/Gateway/{Peer,Gateway}/Options.{h,cpp}: Added a missing - #define ACE_BUILD_DLL and a missing ACE_Svc_Export. Thanks to - Jeff Hellzen <JHellzen@hwdcsaws.cahwnet.gov> for reporting this. - - * ace/Select_Reactor.cpp (suspend_i): The clr_bit call was - mistakenly made on the write mask, rather than the read mask. - Thanks to Billy Quinn <ace-users@lads.com> for reporting this. - -Thu Jan 08 16:02:33 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * ace/Hash_Map_Manager.{h,cpp}: Added STL-styled iterator - functions. - - * tests/Map_Manager_Test.cpp: - * tests/Hash_Map_Manager_Test.cpp: - * tests/Conn_Test.cpp: - * ace/Local_Name_Space.cpp: - * ace/Filecache.cpp: Added missing template instantiation codes. - -Thu Jan 08 11:05:15 1998 <coryan@CHA-CHA> - - * ace/ace_dll.dsp: - * ace/ace_lib.dsp: - Added the new Shared_Memory.cpp file to the project. - -Wed Jan 07 20:40:45 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: protect #define of AF_FILE, because Linux/RedHat 5.0 - #defines it in /usr/include/socketbits.h. Thanks to James CE - Johnson <jcej@lads.com> for reporting this. - - * tests/test_config.h (set_output): "if (! test_dir)" instead of - "if (test_dir)" so the output filename ends up non-null. - - * ace/Shared_Memory.{h,cpp},Makefile (~ACE_Shared_Memory): made - virtual and non-line. Added file Shared_Memory.cpp to contain - the definition. - -Wed Jan 07 17:59:57 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * ace/Free_List.i (add): Only delete freeed item when we hit the - high water mark, not the other way around. Thanks to Stuart - Myles <smyles@wsj.dowjones.com> for pointing this out. - -Wed Jan 7 17:46:09 1998 James C Hu <jxh@cs.wustl.edu> - - * examples/Timer_Queue/Thread_Timer_Queue_Test.h: - * examples/Timer_Queue/Thread_Timer_Queue_Test.cpp: - * examples/Timer_Queue/Reactor_Timer_Queue_Test.h: - * examples/Timer_Queue/Reactor_Timer_Queue_Test.cpp: - * examples/Timer_Queue/Driver.h: - * examples/Timer_Queue/Driver.cpp: - Added virtual destructors to classes that needed them. - -Wed Jan 7 16:28:16 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Event_Handler_T.cpp: Reordered the initializers in the - ACE_Event_Handler_T constructor so G++ doesn't complain. Thanks - to Oleg Krivosheev <kriol@fnal.gov> for reporting this. - - * ace/Shared_Memory.h: Added a virtual destructor to the - ACE_Shared_Memory abstract base class so that EGCS won't - complain. Thanks to Oleg Krivosheev <kriol@fnal.gov> for - reporting this. - - * ace/Event_Handler_T.h: Had to include the *.cpp file into .h - when ACE_TEMPLATES_REQUIRE_SOURCE is defined. Thanks to Oleg - Krivosheev <kriol@fnal.gov> for reporting this. - -Wed Jan 07 14:31:34 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.16, released Wed Jan 07 14:31:34 1998. - -Wed Jan 7 13:38:28 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * tests/test_config.h (set_output): Added an environment variable - to specify where to put the log directory: ACE_TEST_DIR. Thanks - to Keith Nicewarner <knicewar@sanders.com> for reporting this. - - * ace/Service_Repository.cpp (close): Removed a fragment of - redundant code. Thanks to Eric Newton <ecn@smart.net> and Alex - Chan <Alex.Chan@Aspect.com> for reporting this. - -Tue Jan 06 17:19:13 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * ace/Containers.{h,cpp} (ACE_Unbounded_Set[_Iterator]): Added - STL-styled iterator functions. Changed the internal - representation of ACE_Unbounded_Set within iterators from a - reference to a pointer. We need this to compare if two - iterators are refering to the same set or not. - -Tue Jan 6 14:11:54 1998 Carlos O'Ryan <coryan@swarm.cs.wustl.edu> - - * tests/Reader_Writer_Test.cpp: - Reverted to revision 4.21 since the latest one (4.23) fails; - thanks to Irfan for this solution. - -Tue Jan 06 10:37:49 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.i (rw_trywrlock_upgrade): moved the #if 0 up a bit so - that it builds on Linux/non-glibc. For some reason, its - preprocessor was getting confused. - - * ace/OS.i (nanosleep): convert nanoseconds to microseconds - by dividing by 1000, not multiplying. Thanks to - Mitch Kuninsky <Mitch_Kuninsky@avid.com> for reporting this. - - * ace/config-mvs.h: added ACE_LACKS_LONGLONG_T because MVS doesn't - support it. Thanks to Chuck Gehr <gehr@sweng.stortek.com> for - confirming that. - -Mon Jan 5 01:20:51 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * ace/Map_Manager.{h,cpp}: Made operator == and != and their - arguments const. - -Sun Jan 04 19:45:47 1998 David L. Levine <levine@cs.wustl.edu> - - * ace/IPC_SAP.cpp (enable,disable): moved return 0 at end - to inside the #endif, to avoid warning about unreachable - return statement on ghs/VxWorks. Thanks to Brian Mendel - for reporting this. - - * ace/OS.i (rw_trywrlock_upgrade): moved return 0 inside the - #if 0 block to avoid warning about unreachable return statement - on ghs/VxWorks. Thanks to Brian Mendel for reporting this. - - * ace/OS.i (rw_trywrlock_upgrade): added ACE_UNUSED_ARG (rw) - to temporary code. - - * ace/Basic_Types.{h,i},README: replaced ACE_HAS_LONGLONG_T with - ACE_LACKS_LONGLONG_T. ACE_HAS_LONGLONG_T was no longer - being used on many platforms with the scheme in Basic_Types.h. - From now on, we'll assume that a platform has long long - (unless ACE_LACKS_LONGLONG_T is #defined), which should help - porting to new platforms. - - * ace/config-aix-4.{1,2}.x.h,config-dgux-4.11-epc.h, - config-linux-{common,kcc}.h,config-lynxos.h,config-m88k.h, - config-osf1-*.h,config-sunos5.*.h,config-tandem.h: - removed ACE_HAS_LONGLONG_T. - - * ace/config-chorus.h,config-psosim-g++.h,config-vxworks*.h: - added ACE_LACKS_LONGLONG_T. - - * ace/Basic_Types.{h,i} (ACE_U_LongLong): fixed operator/, and - added operator% and operator--. - - * ace/High_Res_Timer.*: changed type of global_scale_factor_ from - u_long to ACE_UINT32. Removed some hacks now that ACE_U_LongLong - has operator%. - - * tests/Time_Value_Test.cpp: added tests of ACE_U_LongLong - operator/ and operator%. - - * ACE-INSTALL.html: added link to Cygus repo patches. Thanks - to Achint Sandhu <sandhu@nortel.ca> for providing the URL. - - * ace/OS.{h,i}: added ACE_OS::putenv (). - - * tests/Env_Value_Test.cpp: use ACE_OS::putenv (). - - * tests/Env_Value_Test.cpp: expect (unsigned long) -10 for - TEST_VALUE_NEGATIVE converted to an unsigned long. - Thanks to Jack Erickson <JErickson@exchange.ml.com> for - verifying that it _should_ return (unsigned long) -10. - - * ace/OS.i (strtoul): on Linux with glibc, use ::strtol instead of - ::strtoul and cast the result to an unsigned long. Its ::strtoul - always seems to return 0 for a negative number, at least on - RedHat 4.2/Linux 2.0.30/Alpha with glibc 2.0 (0.961212-5). - - * ace/config-linux-common.h: factored out - ACE_LACKS_SOME_POSIX_PROTOTYPES from glibc and non-glibc - sections to common section. - - * etc/purify.solaris2: added suppression of MIU from putenv. - -Sun Jan 04 17:19:31 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.15, released Sun Jan 04 17:19:31 1998. - -Sun Jan 4 10:11:53 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS.i (rw_trywrlock_upgrade): Temporarily comments out this - method since we need to finish implementing it correctly. - - * ace/OS.i (rw_trywrlock_upgrade): Fixed a minor bug where we - weren't releasing the lock correctly if there were competing - readers. - - * apps/Gateway/Gateway/Event_Channel.cpp (put): Uncommented the - code that sets the appropriate locking strategy for the Message - Blocks. - - * apps/Gateway/Gateway/Connection_Handler.h: Made the - remote_addr() and local_addr() accessor methods return - const-correct values. - - * build/SunOS5.5/ace/OS: Revised a bunch of #ifdefs so that we can - selectively enable the ACE rwlock emulation, even if we're - compiling with ACE_HAS_STHREADS in order to get the native - Solaris semaphores. - - * apps/Gateway/Peer/Makefile: Tweaked the Makefile rules to fix - some minor problems. Thanks to David Levine for showing the - way... - -Sun Jan 04 12:37:02 1998 David L. Levine <levine@cs.wustl.edu> - - * tests/Env_Value_Test.cpp: hacked to work with - ACE_HAS_NONSTATIC_OBJECT_MANAGER. With that, it uses ::putenv (), - which we should wrap in ACE_OS. Also, expanded TEST_THIS to - provide a more helpful log message than just the ASSERT output. - - * ace/config-vxworks5.x-g++.h: removed - ACE_HAS_TEMPLATE_SPECIALIZATION. - - * tests/Hash_Map_Manager_Test.cpp (main): added casts of i->ext_id_ - and i->int_id_ to (char *) to force g++/VxWorks to apply the - conversion from Dumb_String to char *. - -Sun Jan 04 09:51:59 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * apps/Gateway/Gateway/Options.{h,cpp}: Removed macros to export - this class out of DLL. They are no longer needed. - - * apps/Gateway/Gateway/Gateway.dsp: Updated source file listings. - -Sat Jan 3 22:25:15 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * apps/Gateway/Gateway: Changed the name of the ACE_Event_Channel - class to Event_Channel since this class wasn't a part of the ACE - library. - - * apps/Gateway/Gateway/Event_Channel.h: Updated the inheritance - relationship for Event_Channel so that it inherits from - ACE_Event_Handler rather than ACE_Task. ACE_Task was - overkill... - - * apps/Gateway/Gateway: Removed the supplier_id from all the - files. We didn't use this anyway and it was just confusing. - Also, renamed Proxy_* to Connection_* to underscore the fact - that the Supplier_Handler and Consumer_Handler work atop - *connections*. The previous naming scheme was too abstract for - its own good. - -Sat Jan 03 14:16:40 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * ace/Map_Manager.cpp (ACE_Map[_Reverse]_Iterator): Fixed the - initial iterator position. - -Fri Jan 02 18:43:14 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.14, released Fri Jan 02 18:43:14 1998. - -Fri Jan 2 17:39:04 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * apps/Gateway/Peer/Options.cpp (Options): Rearranged the order of - the parameters in the base/member initialization section to make - GCC happy. Thanks to David Levine for getting on me about this. - - * apps/Gateway/Gateway/Options.h: Changed DEFAULT_TIMEOUT - to MAX_TIMEOUT to be consistent. Thanks to David - Levine for reporting this. - -Fri Jan 2 17:25:05 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * tests/Map_Manager_Test.cpp: Added codes to test STL-styled - functions in ACE_Map_Manager. - - * apps/Gateway/Gateway/Proxy_Handler.cpp: - * apps/Gateway/Gateway/Proxy_Handler_Connector.cpp: - * apps/Gateway/Gateway/Event_Channel.cpp: - * apps/Gateway/Peer/Peer.cpp: - * ace/Local_Name_Space.cpp: - * examples/ASX/Event_Server/Event_Server/Peer_Router.cpp: - * examples/ASX/Event_Server/Transceiver/transceiver.cpp: - * examples/ASX/UPIPE_Event_Server/Supplier_Router.cpp: - * examples/ASX/UPIPE_Event_Server/Consumer_Router.cpp: - * examples/Connection/blocking/SPIPE-connector.cpp: - * examples/Connection/misc/test_upipe.cpp: - * examples/Connection/non_blocking/test_sock_connector.cpp: - * examples/Connection/non_blocking/test_spipe_connector.cpp: - * examples/Connection/non_blocking/test_tli_connector.cpp: - * examples/Connection/non_blocking/test_lsock_connector.cpp: - * netsvcs/clients/Tokens/manual/manual.cpp: - * netsvcs/lib/TS_Clerk_Handler.cpp: - * apps/Gateway/Gateway/Proxy_Handler_Connector.cpp: Added some - more explicit template instantiations. - - * ace/Map_Manager.{h,cpp}: Added back comparison operators for - ACE_Map_Iterator and ACE_Map_Reverse_Iterator. - - * ace/Token_Invariants.cpp: - * ace/Token_Collection.cpp: - * ace/Token_Manager.cpp: Added missing ACE_Map_Reverse_Iterator. - - * ace/Map_Manager.h: Remove unnecessary method declarations from - ACE_Map_Iterator adn ACE_Map_Reverse_Iterator. - - * examples/Connection/non_blocking/test_tli_connector.cpp: - * examples/Connection/non_blocking/test_sock_connector.cpp: - * examples/Connection/non_blocking/test_lsock_connector.cpp: - * examples/Connection/non_blocking/test_spipe_connector.cpp: - * examples/Connection/misc/test_upipe.cpp: - * examples/Connection/blocking/SPIPE-connector.cpp: - * examples/ASX/UPIPE_Event_Server/Supplier_Router.cpp: - * examples/ASX/UPIPE_Event_Server/Consumer_Router.cpp: - * examples/ASX/Event_Server/Transceiver/transceiver.cpp: - * examples/ASX/Event_Server/Event_Server/Peer_Router.cpp: - * apps/JAWS/clients/Caching/http_handler.cpp: - * apps/JAWS/clients/Blobby/Blob.cpp: - * apps/Gateway/Peer/Peer.cpp: - * apps/Gateway/Gateway/Event_Channel.cpp: - * apps/Gateway/Gateway/Proxy_Handler.cpp: - * apps/Gateway/Gateway/Proxy_Handler_Connector.cpp: - * tests/Reactor_Performance_Test.cpp: - * tests/Priority_Reactor_Test.cpp: - * tests/Map_Manager_Test.cpp: - * tests/Conn_Test.cpp: - * netsvcs/lib/TS_Clerk_Handler.cpp: - * ace/Token_Manager.cpp: - * ace/Token_Invariants.cpp: - * ace/Token_Collection.cpp: - * ace/Local_Name_Space.cpp: Added explicit template instantiation - for ACE_Map_Iterator_Base. - - * ace/Map_Manager.{h,cpp}: Added a new class ACE_Map_Iterator_Base - to abstract out most common functionalities from - ACE_Map_Iterator and ACE_Map_Reverse_Iterator. Added some - missing functions to both iterator classes. - -Fri Jan 2 12:39:45 1998 Carlos O'Ryan <coryan@swarm.cs.wustl.edu> - - * ace/Map_Manager.h: - Another declaration removed from ACE_Map_Reverse_Iterator, - operator* is defined in the base class. - - * tests/Conn_Test.cpp: - * tests/Priority_Reactor_Test.cpp: - * tests/Reactor_Performance_Test.cpp: - The instantiation of ACE_Map_Manager *requires* instantiating - ACE_Map_Reverse_Iterator. - - * ace/OS.i: - Cannot use ACE_ASSERT in OS.i, at least it make SGI compiler - unhappy. - - * ace/Map_Manager.cpp: - Several methods for ACE_Map_Manager were not given their fully - qualified name. - - * ace/OS.i: - The error code should be stored in errno, not in "error". - -Fri Jan 2 17:02:36 1998 Chris Cleeland <cleeland@tango.cs.wustl.edu> - - * tests/run_tests.sh: Added Env_Value_Test to the list of tests to - run. - - * tests/Env_Value_Test.cpp (main): Fixed wrong test for negative - values with unsigned longs. Thanks to David Levine and - Linux/Alpha for finding this. - - * tests/Env_Value_Test.cpp (main): Changed the definition of the - TEST_THIS macro to be all on one line. While ugly, this helps in - locating the line when ACE_ASSERT fails, since allowing the macro - to have multiple lines skews the line # reported by ACE_ASSERT. - - * ace/Env_Value_T.*: Fixed lots of little bugs that got past both - g++ and Solaris compilers, but were caught by the ever-skeptical - SGI compiler. This included providing a definition for the missing - 'varname(void)' method. - -Fri Jan 02 10:12:28 1998 Steve Huston <shuston@riverace.com> - - * ace/config-hpux-10.x-g++.h: Added #define HPUX_10 to catch OS - version-dependent code correctly. - -Fri Jan 02 07:06:44 1998 David L. Levine <levine@cs.wustl.edu> - - * apps/Gateway/Gateway/Makefile: restored the .obj/*.o targets. - They don't affect the build of the shared lib and are used if - the unshared lib is built. - - * apps/Gateway/Gateway/Options.h: added missing , in enum definition. - - * apps/Gateway/Gateway/Options.cpp (Options ctor): reordered - initializers to match declaration order. - - * apps/Gateway/Gateway/Config_Files.cpp (read_entry): fixed syntax - error, missing "if" after "else". Added #include of Options.h. - - * ace/OS.i (rw_trywrlock_upgrade): added ACE_UNUSED_ARG (rw) with - STHREADS. - - * ace/Synch.i (tryacquire_write_upgrade): only define if - ACE_HAS_THREADS. - - * tests/test_config.h: hacked ACE_INIT_LOG to work on VxWorks. - VxWorks gets upset if its asked to unlink a nonexistant file. - - * ace/Basic_Types.*: if using ACE_U_LongLong, #define - ACE_USES_ACE_U_LONGLONG and use that in the .i and .cpp files - instead of repeating all of the complex preprocessor tests - in Basic_Types.h. Thanks to Carlos for suggesting this. - - * ace/Env_Value_T.cpp: added protection against multiple inclusion. - - * tests/Env_Value_Test.cpp: added ACE_Env_Value<long> instantiation. - -Fri Jan 02 00:47:33 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.13, released Fri Jan 02 00:47:33 1998. - -Fri Jan 2 00:17:04 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * Reader_Writer_Test.cpp: Changed from an ACE_RW_Mutex to an - ACE_RW_Thread_Mutex so that we can use the new - tryacquire_write_upgrade() method. - - * ace/Env_Value_T.h: Reformatted the file a bit to conform to - ACE coding conventions. - -Thu Jan 1 12:04:29 1998 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * apps/Gateway/Peer/Options.cpp (parse_args): Changed the option - to set the passive acceptor port from -p to -a. This is now - consistent with the option for active connector port number, - which is -c. - - * apps/Gateway/Gateway/Gateway.cpp (fini): Added a call to - ACE::remove_stdin_handler(), which removes the handler for - ACE_STDIN so the gatewayd will shutdown cleanly. - - * ace/ACE: Added a new method, remove_stdin_handler(), which is - the inverse of the register_stdin_handler() method. However, - I'm not sure how to make this work on NT! - - * examples/Reactor/Misc/notification.cpp: Removed some - CHORUS-specific, which should be a no-op with our new signal() - and sigaction() scheme. - - * ace/OS.i: Enhanced the signal() and sigaction() wrappers so that - they simply ignore signums of 0. This means that we don't need - to special case all this stuff throughout ACE anymore. - - * ace/OS.h: Changed all the #defines for various SIG* so that they - are all 0. This way, we can ensure that the signal() functions - in ACE_OS can correctly ignore these signals. This should clean - up a ton of platform-dependent code throughout ACE. - - * apps/Gateway/Gateway/File_Parser.cpp (getint): Added a new - special-case check for the '*' character, which indicates "use - the default value for this field." - - * apps/Gateway/Gateway/Options: Added a new option to provide a - default max retry timeout. - - * tests/Reader_Writer_Test.cpp: Enhanced the test so that it - exercises the new "upgrade" feature of RW_Thread_Mutexes. - - * ace/{OS,Synch}: Added new methods, - ACE_OS::rw_trywrlock_upgrade() and - ACE_RW_Thread_Mutex::tryacquire_write_upgrade() that - conditionally upgrade a read lock to a write lock. This only - works if there are no other readers present, in which case the - method returns 0. Otherwise, the method returns -1 and sets - errno to EBUSY. Note that the caller of this method *must* - already possess this lock as a read lock. - - * apps/Gateway/{Gateway,Peer}/Makefile: Removed the $(VLIB) macro - so that we don't try to build the lib*.a files by default. - Thanks to David Levine for explaining this. - - * apps/Gateway/Gateway: Added new Options.{h,cpp} files and - rearranged the program to centralize all options processing in - an Options Singleton. - - * apps/Gateway/Peer/Options.cpp (parse_args): Changed the - connector and acceptor options to use a more generalizable - format. Now we can set the Consumer and Supplier port numbers - independently. - - * apps/Gateway/Gateway/Event.h: This is currently where the macros - from the following bullets have been moved, so they can be - shared by the Peer and Gateway components. - - * ace/OS.h: Moved the ACE_DEFAULT_{GATEWAY,PEER}_SERVER_PORT - macros into the Gateway application, where they belong. - - * apps/Gateway/Peer/Options.cpp (parse_args): Changed the option - to set the passive acceptor port from -p to -a. This is now - consistent with the option for active connector port number, - which is -c. - - * apps/Gateway/Gateway/Gateway.cpp (fini): Added a call to - ACE::remove_stdin_handler(), which removes the handler for - ACE_STDIN so the gatewayd will shutdown cleanly. - - * ace/ACE: Added a new method, remove_stdin_handler(), which is - the inverse of the register_stdin_handler() method. However, - I'm not sure how to make this work on NT! - - * examples/Reactor/Misc/notification.cpp: Removed some CHORUS-specific - code, which should be a no-op with our new signal() and sigaction() - scheme. - - * ace/OS.i: Enhanced the signal() and sigaction() wrappers so that - they simply ignore signums of 0. This means that we don't need - to special case all this stuff throughout ACE anymore. - - * ace/OS.h: Changed all the #defines for various SIG* so that they - are all 0. This way, we can ensure that the signal() functions - in ACE_OS can correctly ignore these signals. This should clean - up a ton of platform-dependent code throughout ACE. - - * apps/Gateway/Gateway/Gateway.cpp (init): Revised the code a bit - to factor out platform-dependent signal handling stuff in the - right place. - - * apps/Gateway/Gateway/gatewayd.cpp: Cleaned up the gatewayd main - program so that it reuses more ACE components and so that it is - more portable. - -Thu Jan 01 18:24:56 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * apps/Gateway/Gateway/Gateway.dsp: Added new file Options.cpp. - - * apps/Gateway/Gateway/Options.cpp (ACE_BUILD_SVC_DLL): Added - #define ACE_BUILD_SVC_DLL. - - * netsvcs/lib/Base_Optimizer.i: Removed this file. - - * netsvcs/lib/Base_Optimizer.h: Added codes to include in the - source file conditionally if needed. - - * netsvcs/lib/Base_Optimizer.cpp: Moved template implementation - from .i file. Removed ACE_INLINE keyword. - - * netsvcs/lib/Server_Logging_Handler.cpp: Added explicit template - instantiation macros. - -Thu Jan 01 17:13:06 1998 David L. Levine <levine@cs.wustl.edu> - - * apps/Gateway/Peer/Peer.cpp (Peer_Connector::open): added - return 0 statement at end. - - * apps/Gateway/Gateway/Options.cpp (Options ctor): reordered - initializers to match declaration order. - - * apps/Gateway/Gateway/Makefile: removed LOBJ and SHOBJ macros - because they're not needed. And, restored BUILD so that static - lib will be built if requested. - - * include/makeinclude/rules.lib.GNU: added OBJEXT macro, defaulted - to "o", and use it in VOBJS definition. - - * include/makeinclude/wrapper_macros.GNU: if building shared libs - only, set OBJEXT to "so" and VDIR to .shobj/, so that only shared - objects will be built. - - * apps/Gateway/Peer/Peer.cpp: added template instantiations. - - * netsvcs/lib/Base_Optimizer.*,Makefile: moved inline functions - to .i file and changed inline to ACE_INLINE. - - * netsvcs/lib/Server_Logging_Handler_T.{h,cpp}: don't use - Base_Optimizer with g++. Some versions of g++ 2.7.x have - trouble with its inline template member functions, especially - if __ACE_INLINE__ is disabled. - -Thu Jan 01 07:56:17 1998 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.12, released Thu Jan 01 07:56:17 1998. - -Thu Jan 01 00:28:46 1998 Nanbor Wang <nw1@cs.wustl.edu> - - * apps/Gateway/Peer/Peer.dsp: Added a new file "Options.cpp" into - this project. - - * tests/Env_Value_Test.cpp (main): Changed variable name: 'environ' - to 'envp'. Real OSes don't understand what the word means - and Access Violations will occur. - - * tests/run_tests.bat: Added Env_Value_Test. - - * tests/version_tests/version_tests.dsw: - * tests/tests.dsw: Added Env_Value_Test.dsp. - - * tests/version_tests/Env_Value_Test.dsp: - * tests/Env_Value_Test.dsp: New file. - -Thu Jan 1 00:00:00 1998 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * Happy New Year! Here's a thought for the new year, courtesy of - Sumedh: - - Nothing in the world is permanent, - and we're foolish when we ask anything to last; - but surely we're still more foolish - not to take delight in it while we have it. - - - W.Somerset Maugham (Razor's Edge) - -Wed Dec 31 19:10:59 1997 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * apps/Gateway/Peer/Peer: Added the hooks for making the Peer a - Connector, as well as an Acceptor. Now it's time to have some - fun. - - * apps/Gateway/Peer/peerd.cpp: Cleaned up the peerd main program - so that it reuses more ACE components and so that it is more - portable. - - * netsvcs/servers/main.cpp (main): Revised the code a bit to - factor out platform-dependent signal handling stuff in the right - place. - - * apps/Gateway/Peer: Factored out some common initialization code - to create an Options Singleton. Yowza! - - * apps/Gateway/README: Improved the documentation so that its - easier to tell what's supposed to happen and how to configure - the system. - - * apps/Gateway/Peer/Peer: Moved the Peer_Acceptor into the Peer.h - file and updated the documentation to clarify what's going on - better. - - * examples/Reactor/Dgram/: Enhanced the examples to make sure they - close() the dgram endpoint when the object is removed from the - Reactor. Thanks to Evgeny Beskrovny - <evgeny_beskrovny@icomverse.com> for reporting this. - - * ace/Obstack.cpp: Fixed a stupid bug in the ACE_Obstack - constructor. Thanks to David Levine for reporting this. - - * ace/Local_Name_Space_T.h: Updated the documentation for class - ACE_Name_Space_Map to make it more clear. - -Wed Dec 31 21:21:50 1997 David L. Levine <levine@cs.wustl.edu> - - * examples/Reactor/Dgram/{CO,Dgram.cpp} (handle_close): added - ACE_UNUSED_ARG (handle). - - * tests/Env_Value_Test.cpp (main): removed unused arg argv and - local envargv. undef'ed ACE_HAS_NONSTATIC_OBJECT_MANAGER because - it can't deal with the three arguments to main (). - -Wed Dec 31 16:25:16 1997 Chris Cleeland <cleeland@cs.wustl.edu> - - * ace/Env_Value_T.*: Added a new template (ACE_Env_Value) which is - a typesafe encapsulation of environment variables with the - ability to provide a value to be used as a default if the - variable isn't set. An example of its use can be seen in the - corresponding test, tests/Env_Value_Test.cpp. Thanks to Carlos - O'Ryan for giving me the seed that started this. - - * tests/Env_Value_Test.cpp: Added test for the new template added - above. - - * tests/Makefile: Added new Env_Value_Test test to the Makefile. - - * ace/OS.{h,i} (strtod): Added new function for converting strings - to doubles. - -Tue Dec 30 22:07:44 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/High_Res_Timer.cpp (elapsed_time): added a cast in - calculation of useconds to avoid MSVC++ warning about narrowing. - Thanks to Nanbor for reporting this. - - * ace/OS.i (gethrtime): commented out high-res tick counter on Linux - because it doesn't seem to work properly. Use ::gettimeofday () - instead. - - * tests/Basic_Types_Test.cpp (main): replaced #ifdef VXWORKS && ghs - with !ifdef ACE_HAS_WCHAR_TYPEDEFS_CHAR. - - * ace/Map_Manager.cpp (operator*): fixed typo in "ACE_Map_Entry". - - * ace/OS.h: added ACE_NSIG. In addition, I removed this: - - // Why is this defined? It must be a std C library symbol. - #if !defined (NSIG) - #define NSIG 0 - #endif /* NSIG */ - - We #define NSIG elsewhere on platforms (VxWorks, MVS) that need it. - #defining it to be 0 isn't good; it's better to catch the lack of - NSIG at compile time than at run time. - - * ace/{Log_Msg.cpp,OS.i,Select_Reactor.cpp,Signal.*,WFMO_Reactor.i}, - tests/Sigset_Ops_Test.cpp: replace NSIG with ACE_NSIG. - - * ace/Basic_Types.{h,i}: only use u_longlong_t on sun. Added - operator++ to ACE_U_LongLong, and tweaked calculation in operator/. - - * tests/Time_Value_Test.cpp (test_ace_u_longlong_): added test of - operator+, and one simple test of operator/. - - * ace/Filecache.h: removed extra , in enum ACE_Filecache_Flag definition. - -Tue Dec 30 20:32:03 1997 Nanbor Wang <nw1@cs.wustl.edu> - - * ace/Map_Manager.{h,cpp}: Added STL-styled iterator function. - Changed the references to ACE_Map_Manager within interator - classes to pointers so we can check errors. - -Tue Dec 30 14:56:18 1997 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Obstack: Enhanced ACE_Obstack so that it can be given a - pointer to an allocator strategy. This allows users to take - complete control of where the memory comes from for the Obstack. - - * ace/Obstack: Changed the default value of "size" so that it is - (4096 - sizeof (ACE_Obstack)). - - * ace/Obstack: Added a new ACE_Obchunk constructor to simplify - some of the code. - -Mon Dec 29 20:21:32 1997 Steve Huston <shuston@riverace.com> - - * include/makeinclude/rules.lib.GNU: Adjusted dependencies on building - a shared library to be on the object modules it will link in, not - necessarily $(VSJOBJ). - -Mon Dec 29 18:53:35 1997 James C Hu <jxh@cs.wustl.edu> - - * ace/Filecache.h: Added an enum type, ACE_Filecache_Flag, which - gets passed into the ACE_Filecache_Handle constuctor to help - determine whether or not to use memory mapping (on Windows NT, - we sometimes do not want to memory map). - - * ace/Filecache.cpp: Required modifications to accomodate the - ACE_Filecache_Flag parameter. - -Mon Dec 29 15:37:24 1997 Nanbor Wang <nw1@cs.wustl.edu> - - * ace/OS.i (event_timedwait): Added codes to guard against - negative wait time since we are passing an absolute time value - here, we may way past the target time when it gets called. - Thanks to Norbert Rapp <NRapp@nexus-informatics.de> for pointing - this. - -Mon Dec 29 13:57:43 1997 Carlos O'Ryan <coryan@macarena.cs.wustl.edu> - - * ace/XtReactor.cpp: - Fixed syntax error, apparently some kind of typo. - - * ace/OS.h: - The ACE_*_cast() macros had (IMHO) surprizing priority, added - some parenthesis to make them behave as the new cast operators - are supposed to do. - -Mon Dec 29 08:59:28 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/Basic_Types.h: Linux w/o glibc 2.0 doesn't have u_longlong_t, - so use "unsigned long long" for it. - - * ace/Task.cpp (activate): added ACE_UNUSED_ARG (thread_handles) on - non-threaded platforms. - - * ace/Thread_Manager.cpp (wait): added ACE_UNUSED_ARG - (abandon_detached_threads) on non-threaded platforms. - - * performance-tests/Misc/childbirth_time.cpp (prof_ace_os_thread): - added ACE_UNUSED_ARG (iteration) on non-threaded platforms. - - * performance-tests/Misc/preempt.cpp (main): added ACE_UNUSED_ARG - for argc and argv on non-threaded platforms. - - * include/makeinclude/platform_linux.GNU: enabled compiler warnings - by replacing -w with -Wall -Wpointer-arith in CFLAGS. - - * ace/ACE.cpp (count_interface,get_handle,get_ip_interfaces): use - same code for LynxOS as for __unix. - - * ace/OS.h: on LynxOS, redefine NSIG to be one greater than the - highest-numbered signal. LynxOS' signal.h #defines it to be - equal to the highest-numbered signal, 64. - -Mon Dec 29 08:35:17 1997 Steve Huston <shuston@riverace.com> - - * include/makeinclude/platform_hpux_aCC.GNU: Removed the PRELIB - and PRELIB_USES_OBJ_ONLY stuff; merged in macros for building - on HP-UX 11.00. - * include/makeinclude/platform_hpux11_aCC.GNU - Removed this file. - The 10.x and 11.x builds are now both controlled by - platform_hpux_aCC.GNU. - -Sun Dec 28 10:50:54 1997 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.11, released Sun Dec 28 10:50:54 1997. - -Sat Dec 27 18:03:52 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * ace/Log_Msg.cpp (dump): Sould test if we have a valid thr_desc - before dumping the information of it. Thanks to Luca Priorelli - <lucapri@mbox.vol.it> for pointing this out. - -Sat Dec 27 16:13:59 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/{OS.i,config-linux-common.h,README} (getrusage,{get,set}rlimit): - added ACE_HAS_RUSAGE_WHO_ENUM and ACE_HAS_RLIMIT_RESOURCE_ENUM - support. glibc 2.0 for Linux uses enums for the first arguments - to ::getrusage and ::{get,set}rlimit. - -Fri Dec 26 18:04:06 1997 <nw1@JIG> - - * ace/Containers.h: Added iterator traits for following classes. - ACE_Unbounded_Set - ACE_Unbounded_Queue - ACE_Unbounded_Stack - ACE_Double_Linked_List - ACE_Fixed_Set - ACE_Bounded_Set - Thanks to <???> for recommending this. - -Fri Dec 26 12:03:05 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/config-osf1-4.0.h: added support for DCE threads, and merged in - config-osf1-4.0-g++.h. Thanks to Frederic Andres - <andres@rd.nacsis.ac.jp> for the DCE threads support #defines. - - * ace/config-osf1-4.0-g++.h: removed. All Digital Unix 4.0 configs - are now supported by config-osf1-4.0.h. - - * ace/platform_osf1_4.0.GNU: added comment on how to use DCE threads - instead of POSIX threads. - - * ace/OS.i (thr_{get,set}prio): don't apply the - !defined(ACE_LACKS_SETSCHED) test if ACE_HAS_DCETHREADS. - - * ace/OS.cpp (thr_create): added || ACE_HAS_DCETHREADS to - #ifndef ACE_LACKS_SETSCHED, so that priority is set with DCETHREADS. - - * ace/OS.cpp (sched_params): moved paren so that DCETHREADS block - will be executed even if ACE_LACKS_SETSCHED. - - * ace/Sched_Params.cpp ({next,previous)_priority): added - ACE_HAS_DCETHREADS check before !defined(ACE_LACKS_SETSCHED). - - * ace/OS.i (cond_timedwait): wrapped first arg to ACE_OSCALL in - parens because it confused cxx. - - * examples/System_V_IPC/SV_Message_Queues/{,T}MQ_Server.cpp: renamed - variable msgque because it conflicted with a symbol in the DEC cxx - w/DCE threads headers. - - * ACE-INSTALL.html: expanded discussion of gcc config.status file. - Thanks to Hans Rohnert <rohnert@scr.siemens.com> for figuring - out the role of the config.status file. - -Fri Dec 26 06:44:25 1997 <nw1@JIG> - - * ace/OS.h (ACE_OS::ace_flock_t): Added a new field "lockname_" to - remember the lock filename. - - * ace/OS.i (flock_init,flock_destroy): We need to remember the - name of the lock file when initializing the lock and unlink the - file when destroying the lock. - -Thu Dec 25 18:08:42 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/XtReactor.cpp: Removed g++ warning about initialization - order, enabled read/write masks for - ACE_Event_Handler::CONNECT_MASK; fixed typo that checks a write - mask, but sets a read mask; removed unused variable "zero." - Thanks to Eric Newton for these fixes. - -Thu Dec 25 17:17:16 1997 <nw1@JIG> - - * ace/config-win32.h: * ace/config-win32-common.h: Added macro - definitions not to use C++-styled castings on Win32 by default - so that we don't depend on the RTTI support. - -Thu Dec 25 16:18:03 1997 <irfan@PACHANGA> - - * ace/SString.h (ACE_CString and ACE_WString): Added c_str() to - make it similar to STL's string class. - -Wed Dec 24 19:59:15 1997 David L. Levine <levine@cs.wustl.edu> - - * tests/IOStream_Test.cpp (client): changed stream error message - from "timeout" to "timed out", so that run_tests.check won't - complain about it. - -Wed Dec 24 18:37:49 1997 <nw1@JIG> - - * ace/Synch.i (get_thread_id,get_nesting_level): Added ACE_INLINE. - -Wed Dec 24 18:31:31 1997 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.10, released Wed Dec 24 18:31:31 1997. - -Wed Dec 24 14:27:33 1997 <nw1@JIG> - - * ace/Synch.{h,i,cpp} (ACE_Recursive_Thread_Mutex): Changed - ACE_Recursive_Thread_Mutex to inherit from ACE_Thread_Mutex on - Win32 because ACE_Thread_Mutex is already recusive on Win32. - -Wed Dec 24 08:14:32 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/Log_Msg.cpp (instance): without threads, dynamically create - the ACE_Log_Msg instance instead of using a static. Now, without - threads, the ACE library has no statics that require destruction - (at least on Solaris/g++ and Linux). And, all of the ACE tests - Purify cleanly. - - * ace/OS.cpp (ACE_Thread_Adapter::invoke): wrapped - rethrow_w32_structural_exception () call with #ifdef ACE_WIN32. - It was eliminated by the ACE_SEH_EXCEPT macro on non-WIN32 - platforms, but some compilers (cxx) warned about the - unreachable code. This call is only supported on WIN32 anyways. - - * ace/OS.h: LynxOS has ACE_HAS_PTHREADS_1003_DOT_1C, but not - ACE_HAS_PTHREAD_SIGMASK. - - * tests/Time_Value_Test.cpp: only test ACE_U_LongLong on Solaris, - and other platforms with STHREADS. - -Wed Dec 24 04:25:49 1997 <nw1@DINGO> - - * ace/OS.i (sema_init): Added checking for initial semaphore - counts and set the semaphore to non-signaled state if the - initial count is zero. Changed the internal variable lock from - a mutex to a thread_mutex (Critical Section.) - -Tue Dec 23 21:06:14 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/TLI_Stream.cpp (get_remote_addr): added ACE_UNUSED_ARG - for the arg if ! ACE_HAS_SVR4_TLI. - - * examples/IPC_SAP/SOCK_SAP/CPP-inserver-poll.cpp (main): declare - local "i" outside the for loop because it is used after the loop. - -Tue Dec 23 19:34:10 1997 <nw1@JIG> - - * ace/OS.i: Added comments for semaphore simulation. - -Tue Dec 23 19:33:59 1997 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.9, released Tue Dec 23 19:33:59 1997. - -Tue Dec 23 19:05:32 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * tests/IOStream_Test.cpp (server): Changed sleep(30) to - sleep(5). This seems more reasonable. Thanks to - James Johnson for this suggestion. - -Tue Dec 23 16:43:14 1997 Steve Huston <shuston@riverace.com> - - * ace/README: Added descriptions for ACE_HAS_PTHREADS_1003_DOT_1C (this - was in use previously but not documented), ACE_HAS_XTI (new; XTI is - the POSIX superset of TLI - this implies ACE_HAS_TLI), - ACE_NEW_THROWS_EXCEPTIONS (new; the compiler throws bad_alloc if - a 'new' memory allocation fails). - - * ace/Hash_Map_Manager.cpp - Workaround for an aC++ bug in the HP-UX 11 - version of the compiler. - - * ace/INET_Addr.cpp: Note the failing host name in the error message - from ACE_INET_Addr::ACE_INET_Addr(u_short, const char []). - - * ace/OS.h: Changes made while porting to HP-UX 11: - - If ACE_HAS_PTHREADS_1003_DOT_1C (POSIX.1C), it implies (by - way of POSIX.1C) that these also be set: - ACE_HAS_PTHREAD_SIGMASK - ACE_HAS_2_PARAM_ASCTIME_R_AND_CTIME_R - This is intended to make the process of configuring a new - new platform easier. - - If the new config macro ACE_NEW_THROWS_EXCEPTIONS is set, include - the file <new>. The "proper" thing to do is to include - <stdexcept>; however, at least on HP-UX 11, this will cause - other things to get brought in, like <memory>, which will - screw up auto_ptr. The day that all this standard library and - STL stuff is finally straight, there will be dancing in the - streets... - - Support including the proper header file for XTI when it's - supposed to use <xti.h> rather than <tiuser.h> (POSIX specifies - XTI, a superset of TLI, to have <xti.h>) - - If ACE_NEW_THROWS_EXCEPTIONS is set, the ACE_NEW and - ACE_NEW_RETURN macros handle the exception. - - * ace/OS.i: Fixed check for failure in ACE_OS::sema_init. - - * ace/config-hpux11.h: - * ace/config-hpux11-hpc++.h: New configs for HP-UX 11.00 - - * include/makeinclude/platform_hpux11_aCC.GNU: Platform macros for - HP-UX 11. It is set up for 32-bit build with comments for how to - do a 64-bit build. - - * examples/Makefile: Added Naming to the directory list to build. - - * examples/IPC_SAP/SOCK_SAP/CPP-inserver-poll.cpp: Changed test from - ACE_HAS_SVR4_POLL to ACE_HAS_POLL. - - * examples/IPC_SAP/SOCK_SAP/C-inclient.cpp: - * examples/IPC_SAP/SOCK_SAP/CPP-unclient.cpp: - * examples/IPC_SAP/SOCK_SAP/CPP-unserver.cpp: - * examples/IPC_SAP/SOCK_SAP/FD-unserver.cpp: Fixed a char * being - initialized by a string literal. - - * examples/Map_Manager/test_hash_map_manager.cpp: Added #include - "ace/SString.h" - - * examples/Mem_Map/IO-tests/test_io.cpp: Fixed a char * being - initialized by a string literal. Fixed diagnostic statements. - - * examples/Naming/test_multiple_contexts.cpp: - * examples/Naming/test_non_existent.cpp: - * examples/Naming/test_writers.cpp: Added Id line for CVS. - - * tests/Time_Value_Test.cpp: Added #include "test_config.h" - -Tue Dec 23 16:12:28 1997 <nw1@DINGO> - - * ace/OS.i: Must use ACE_OS::sema_wait instead of waiting on the - singling event when using simulated semaphores. - -Tue Dec 23 08:32:00 1997 Chris Gill <cdgill@cs.wustl.edu> - - * include/makeinclude/platform_sunos5_g++.GNU - Removed trailing spaces from CC and CXX macros - (the trailing space behind CXX was giving the TAO_IDL - make fits) - -Tue Dec 23 07:26:28 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/TLI.cpp (get_local_addr): return -1, instead of 0, and - set errno (with ACE_NOTSUP_RETURN) if ! ACE_HAS_SVR4_TLI. - ({set,get}_option): added ACE_UNUSED_ARfor all - arguments if ! ACE_HAS_SVR4_TLI. - - * ace/TLI_Stream.cpp (get_remote_addr): return -1, instead of 0, - and set errno (with ACE_NOTSUP_RETURN) if ! ACE_HAS_SVR4_TLI. - Declare and set local "name" only if ACE_HAS_SVR4_TLI. - - * ace/TLI_Acceptor.cpp (handle_asynch_event): initialize local - "event" instead of defining it in the switch condition, to - avoid compiler warning about not using it. - - * examples/IPC_SAP/TLI_SAP/db-server.cpp (lookup_name): removed - unused local variable "len". - - * include/makeinclude/platform_osf1_4.0.GNU: removed some warning - suppressions that are no longer needed. - - * ace/Log_Msg.cpp (~ACE_Log_Msg): clean up the message queue and - allocated strings on non-threaded platforms. - - * tests/test_config.h (ACE_APPEND_LOG): added a close () of the - ace_file_stream before opening it. Forked children inherit the - opened file stream, and fail on their attempt to open it if it - is not explicitly closed. - -Mon Dec 22 22:42:57 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS: Added a few minor changes to Nanbor's Semaphore - emulation for WinCE. - -Mon Dec 22 11:39:19 1997 <irfan@PACHANGA> - - * OS.i (event_timedwait): Added code to handle timeouts. Thanks - to Luca Priorelli <lucapri@mbox.vol.it> for reporting this. - -Mon Dec 22 07:03:13 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/IOStream_Test.cpp (spawn): changed printout from - "Process_Mutex_Test-children" to "IOStream_Test-children". - -Mon Dec 22 03:08:22 1997 <nw1@DINGO> - - * ace/OS.i: Modified ACE_OS::cond_* for WinCE. - - * ace/config-win32-common.h: Prevent ACE from using NT specific - API on Win CE. - -Sun Dec 21 16:19:40 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * ace/OS.{h,i} (ACE_OS::sema_*): Added semaphore simulation code - for WinCE. This code is currently protected - ACE_USES_WINCE_SEMA_SIMULATION for debugging the code on NT. - This flag should soon be obsolete and changed to ACE_HAS_WINCE. - - Moved ACE_event_t macro definition for Win32 before ACE_sema_t - definition for NT. I need to use it in semaphore simulation. - - * ace/OS.h (class ACE_Cleanup_Info::operator ==, !=): Removed - keyword struct. ACE_Cleanup_Info is a class now. - - * ace/Containers.h (class ACE_Double_Linked_List): Removed - unused function declarations. - - * ace/Containers.h (class ACE_DNode): Removed ACE_DNode class. - -Sun Dec 21 18:47:09 1997 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.8, released Sun Dec 21 18:47:09 1997. - -Sun Dec 21 12:39:38 1997 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/OS.h (ACE_OS): Made the netdb_{acquire()/release()} methods - public to help ACE users stuck on weird OS platforms. Thanks to - Arturo Montes <mitosys@colomsat.net.co> for this suggestion. - - * ace: Changed all ACE class documentation so that it puts the - =TITLE and =DESCRIPTION comments *after* the opening '{' rather - than before. I did this since (1) it makes emacs C++ mode - happier and (2) it makes Rational Rose reverse-engineering tool - happier too. Thanks to Thomas Jordan - <Thomas_Jordan@deluxedata.com> for suggesting this. - - * ace/LOCK_SOCK_Acceptor.h: The accept() method of class - ACE_LOCK_SOCK_Acceptor is no longer a virtual. I don't know why - this was made virtual in the first place, but it was a mistake. - -Sun Dec 21 15:24:42 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * ace/OS.h (class ACE_Cleanup_Info::operator ==, !=): Removed - keyword struct. ACE_Cleanup_Info is a class now. - - * ace/Containers.h (class ACE_Double_Linked_List): Removed - unused function declarations. - - * ace/Containers.h (class ACE_DNode): Removed ACE_DNode class. - -Sat Dec 20 11:05:45 1997 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * tests/test_config.h (ACE_INIT_LOG): Added (%P|%t) to an - ACE_DEBUG statement to make it consistent. - - * tests/IOStream_Test.cpp (spawn): Revised the test so that it - prints data correctly to the child processes. - - * examples/Shared_Malloc/test_multiple_mallocs.cpp (main): Moved - the final print statement inside one of the if blocks so that it - appears only every second invocation to remind people to rerun - the program to remove the leaks. Thanks to Tom Arbuckle - <arbuckle@uran.informatik.uni-bonn.de> for suggesting this. - - * ace/Malloc*: Changed ACE_MALLOC_STATS to ACE_HAS_MALLOC_STATS - since this is more consistent with ACE conventions. Also, - added ACE_HAS_MALLOC_STATS to the ace/README file. Thanks - to Tom Arbuckle <arbuckle@uran.informatik.uni-bonn.de> for - suggesting this. - - * ace/Memory_Pool.cpp (ACE_Shared_Memory_Pool): Fixed the logic - for naming the backing store if the sscanf() call fails. Thanks - to Tom Arbuckle <arbuckle@uran.informatik.uni-bonn.de> for - suggesting this. - -Sat Dec 20 15:39:00 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * tests/IOStream_Test.cpp (spawn): Changed ACE_LACKS_EXEC to - ACE_LACKS_FORK. - -Sat Dec 20 09:42:06 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: changed ACE_SEH_FINALLY to be if (1) on Unix, not - just pSOS. - -Fri Dec 19 23:20:42 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * tests/IOStream_Test.cpp (client): Removed an extra read. - - * ace/OS.h: ACE_SEH_FINALLY should always be executed on UNIX - platforms. - -Fri Dec 19 18:49:26 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/IOStream_T.i: Enhanced ACE_IOStream and its companions as - follows: - - - ACE_Streambuf_T<> now distinguishes recv() timeout failure from - a "real" stream failure - - - ACE_IOStream::eof() honors the streambuf's recv_timeout() value - and returns FALSE if the eof() check fails on timeout. - - - IOStream_Test.cpp tests the streambuf timeout via the stream eof() - and retrys if possible. - - Thanks to James Johnson <jcej@lads.com> for these fixes. - - * ace/Log_Record.i: Removed a no-op assignment to Time_Value - in the encode() method. Thanks to Patrick J. McNerthney - <pat@thememedia.com> for reporting this. - - * ace/Log_Record.i: The time stamp put into the ACE_Log_Record - gets all messed up when the ACE_Log_Record is transmitted to the - Server_Logging_Handler and the client is run on a little-endian - architecture. This is because the ACE_Log_Record::encode method - does: - - this->time_stamp_ = ACE_Time_Value - - htonl (this->time_stamp_.sec ()), - htonl (this->time_stamp_.usec ())); - - When the ACE_Time_Value is passed in the resulting values from - "htonl", it calls "normalize" which totally messes with the - values passed. Thanks to Patrick J. McNerthney - <pat@thememedia.com> for reporting this. - - * ace/OS.i: Fixed ctime_r() so that it doesn't crash if a NULL is - returned from ctime_r(). Thanks to Patrick J. McNerthney - <pat@thememedia.com> for reporting this. - -Fri Dec 19 09:44:28 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.{h,i}: fixed sigwait for DEC cxx, yet again for cxx - V6.0-010. cxx #defines sigwait, and was getting confused - with ACE_OS::sigwait. I replaced Thilo's #undef sigwait, - and updated the ::sigwait declaration to what cxx uses internally. - - * examples/IPC_SAP/SOCK_SAP/CPP-inserver.cpp (server): cast arg - to long before casting to ACE_HANDLE, and removed function-style - cast. - -Fri Dec 19 00:39:40 1997 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.7, released Fri Dec 19 00:39:40 1997. - -Thu Dec 18 19:47:47 1997 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * examples/IPC_SAP/SOCK_SAP: Stripped the CPP-in{server,client}.cpp - tests down to their bare essence to only test the core features - of sockets. Thanks to Bryan and Brian for suggesting this. - - * ace/Message_Queue.cpp (ACE_Message_Queue_Iterator): Changed - queue to q to avoid clashes with STL. Thanks to Brian Mendel - <bmendel@mdc.com> for reporting this. - - * ace/OS.h: The server logging daemon is implemented using TCP/IP - under Win32. Therefore, we can simplify the creation of the - ACE_DEFAULT_LOGGER_KEY. Thanks to Patrick J. McNerthney - <pat@thememedia.com> for reporting this problem. - -Thu Dec 18 22:35:45 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: fixed typo in end-of-comment marker. - - * ace/OS.h: on Digital Unix 4.0, only declare sigwait () with cxx, - not with g++. Thanks to Erik Margraf <erik@asgard.gud.siemens.co.at> - for verifying this. - - * examples/IPC_SAP/SOCK_SAP/CPP-in{client,server}.cpp: tweaked - so that they build on VxWorks (spawn needs the (ACE_THR_FUNC) - cast), and without threads. - - * tests/run_tests.vxworks: updated VxWorks test status. - - * ace/Basic_Types.h: look for ULONGLONG_MAX, which Irix 6.x uses. - And, hard-code ACE_SIZEOF_LONG_DOUBLE to be 16 on __sgi, because - its float.h doesn't always allow us to distinguish them. - Thanks to Carlos for reporting these. - - * ace/High_Res_Timer.cpp,Profile_Timer.cpp: removed dependence - on ACE_U_LongLong. Thanks to Carlos for reporting this. - - * tests/Time_Value_Test.cpp: fixed #ifdefs so that ACE_ULongLong - isn't tested on Irix 6.x, etc. Thanks to Carlos for tripping - over this. - - * ace/IOStream_T.i (eof): made the return type, int, explicit. - - * ace/config-linux-common.h: started adding support for threads on - alpha (glibc). - -Thu Dec 18 21:17:46 1997 <nw1@DINGO> - - * ace/ace_dll.dsp: Added project files for Windows CE on SH3 - series CPU and x86 emulator. - - * ace/streams.h: - * ace/OS.h: Added Windows CE flags. - - * ace/OS.cpp (invoke): Added a dummy return to shut gcc up. - - * ace/OS.cpp (invoke): Added a dummy return to shut gcc up. - - * ace/OS.{h,cpp} (rethrow_w32_structural_exception): Added this - function, which prints out an error message to indicate the - occurance of structural exception and rethrow it again. - (ACE_Thread_Adapter::invoke): Rearranged try and except block - for Win32 and added a finally block so when a structural - exception occurs, a thread can clean it up properly, print out - an error message, and rethrow the exception again. - -Thu Dec 18 15:40:50 1997 James C Hu <jxh@cs.wustl.edu> - - * ace/Filecache.{cpp,h}: - - debugged writing a file. - - can select if mapping is desired. - - * ace/OS.h: Added new #defines to facilitate forward declarations - of ACE_Export classes. - -Thu Dec 18 00:15:02 1997 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.6, released Thu Dec 18 00:15:02 1997. - -Wed Dec 17 19:42:38 1997 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * examples/IPC_SAP/SOCK_SAP: Enhanced the - CPP-{inserver,inclient}.cpp tests so that they will spawn - multiple client and server threads to perform the data transfer. - Thanks to Brian Mendel and Bryan Doerr for suggesting this. - - * ace/OS.i: Fixed some very subtle bugs with cond_signal() and - cond_broadcast(). Thanks to Patrick J. McNerthney - <pat@thememedia.com> for reporting these and sending the fixes. - -Wed Dec 17 19:27:06 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * ACE-INSTALL.html (HREF): Added an advise not to use VC5.0 SP3. - - * examples/OS/Process/Makefile: Removed inclusion of - rules.lib.GNU. Thanks to Avi Nash <anash@RedBrick.Com> for - noticing this. - -Tue Dec 16 22:09:42 1997 David L. Levine <levine@cs.wustl.edu> - - * tests/Time_Value_Test.cpp: restored #include test_config.h, - and changed #include ace/config.h to ace/inc_user_config.h. - - * examples/IPC_SAP/TLI_SAP/ftp-server.cpp, - examples/Reactor/Misc/test_timer_queue.cpp, - examples/Service_Configurator/IPC-tests/server/Handle_Timeout.i, - examples/Shared_Malloc/test_malloc.cpp, - examples/Threads/thread_specific.cpp: replaced ACE_HAS_64BIT_LONGS - preprocessor logic with a cast through a long to an int. - A void * is the same size as a long on all currently supported - ACE platforms. The int cast is also there, to avoid compiler - warnings about narrowing. As a side effect, the function-style - casts were removed. - - * ace/config-osf1*.h,README: added comment stating that - ACE_HAS_64BIT_LONGS is deprecated. - - * ace/config-tandem.h: removed commented-out line that mentioned - ACE_HAS_64BIT_LONGS. - - * ace/OS.{h,i} (sigwait): removed DEC cxx version dependence. The - standard two-arg POSIX ::sigwait () seems to work just fine even - on older releases of Digital Unix 4.0, with the sigwait declaration - that we have in OS.h. Thanks to Erik Margraf - <erik@asgard.gud.siemens.co.at> and Thilo - <kielmann@informatik.uni-siegen.de> for helping to sort this out. - - I did change the ACE_OS::sigwait () return value determination - in OS.i. It now depends on whether or not POSIX ::sigwait () - returns 0, to agree with the documenation for POSIX sigwait. - Before, the code checked for a -1 return value from ::sigwait (). - -Tue Dec 16 21:00:49 1997 Chris Gill <cdgill@tango.cs.wustl.edu> - - Single Threaded port of ACE to pSOSim - first checkin - - * psosim/Makefile: (new) make file for new psosim directory - - builds the configured simulator components - - * psosim/drv_conf.c: (new) board driver configuration code for - simulator - - * ace/sys_conf.h: (new) configuration settings for simulator, ACE - - * ace/config-psosim-g++.h: (new) config file for psosim w/ g++ - - * ace/OS.cpp: - * ace/OS.h: - * ace/OS.i: Added psos and psosim specific code - - * ace/Mem_Map.cpp: Removed function style casts which were killing - g++ - - * ace/Message_Block.cpp: Removed function style casts which were - killing g++ - - * ace/Containers.cpp: Removed function style casts which were - killing g++ - - * ace/Map_Manager.cpp: Removed function style casts which were - killing g++ - - * tests/Handle_Set_Test.cpp: Added PSOS #defines, alongside those - for VXWORKS - - * tests/SOCK_Connector_Test.cpp: Added PSOS #defines, alongside - those for VXWORKS - - * tests/Priority_Reactor_Test.cpp: Added PSOS #defines, alongside - those for VXWORKS - - * tests/MM_Shared_Memory_Test.cpp: Added PSOS #defines, alongside - those for VXWORKS - - * tests/Conn_Test.cpp: Added PSOS #defines, alongside those for - VXWORKS - - * tests/run_tests.psosim: (new) script to run only those tests - that are relevant to pSOSim, in its current single threaded - state - - * include/makeinclude/platform_psosim_g++.GNU (new) make macros - for pSOSim using g++ - - -Tue Dec 16 17:23:49 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * tests/Time_Value_Test.cpp: Removed inclusion of "ace/config.h". - - * ace/inc_user_config.h: Added this file to prevent multiple - inclusions of config.h file. Replace `#include "ace/config.h"' - with `#include "ace/inc_user_config.h"'. - - * ace/Log_Msg.cpp: - * ace/OS.h: - * ace/Trace.cpp: - * ace/streams.h: - * ace/iosfwd.h: Replace "config.h" with "inc_user_config.h". - -Tue Dec 16 08:30:41 1997 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * examples/IPC_SAP/FILE_SAP/client.cpp (main): Fixed "lines" to - "links." Thanks to Istvan Buki <ibuki@fedex.com> for reporting - this. - -Tue Dec 16 14:15:48 1997 <irfan@TWOSTEP> - - * ace/OS.i (fstat): Fixed Dr. Schmidt's typo. - -Tue Dec 16 09:17:53 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/Singleton.cpp (cleanup): set instance pointer to 0 after - deleting it. Thanks to Luca Priorelli <lucapri@mbox.vol.it> - for suggesting this. - - * ace/Basic_Types.{h,i}: (ACE_U_LongLong): use ACE_UIN32 instead - of u_long for data members, to ensure that its sizeof is 8 bytes. - - * ace/OS.h: except on WIN32 and Solaris (ACE_HAS_HI_RES_TIMER), - always typedef ACE_hrtime_t to ACE_UINT64. This changes removes - the dependence on ACE_HAS_64BIT_LONGS. - - * ace/OS.cpp (ACE_TSS_Keys): replaced ACE_HAS_64BIT_LONGS - dependence with ACE_SIZEOF_LONG == 8, and similarly for - determination of 4-byte longs. - - * tests/TSS_Test.cpp (worker): replaced ACE_HAS_64BIT_LONGS - preprocessor logic with a cast of the void * argument to a long. - A void * is the same size as a long on all currently supported - ACE platforms. The int cast is also there, to avoid compiler - warnings about narrowing. - - * tests/Time_Value_Test.cpp: replaced use of ! ACE_HAS_64BIT_LONGS - with ACE_SIZEOF_LONG == 4. - -Mon Dec 15 22:47:24 1997 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.5, released Mon Dec 15 22:47:24 1997. - -Mon Dec 15 20:17:10 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS.cpp: Fixed pread() and pwrite() so that they work - correctly even if we don't support threads. Thanks to Irfan for - finding this. - - * ace/OS.i: Tried to fix the fstat() wrapper on Win32 so that - it does the right think for HANDLES. Thanks to Lori Anderson - <lori@probita.com> for pointing this out. - -Mon Dec 15 15:36:07 1997 Nanbor Wang <nw1@cumbia.cs.wustl.edu> - - * ace/Thread_Manager.cpp (exit,wait): Changed these two functions - more for Chorus. Thanks again to Wei Chian - <chiang@tele.nokia.fi> for providing the patch. - -Mon Dec 15 06:42:42 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/Log_Msg.cpp (instance): on non-threaded platforms, only - allocate a new ACE_Log_Msg_message_queue if the (static) pointer - is 0. Thanks to Stephen Coy <stevec@magna.com.au> for - reporting this. - - * ace/config-linux-common.h: updated header comments about - autosensing the platform capabilities. - - * tests/Process_Strategy_Test.cpp: (parse_args): added DEBUG - message if an unsupported concurreny strategy is provided with - a -c option. - -Mon Dec 15 01:12:33 1997 <irfan@TWOSTEP> - - * ace/Synch_T.h (ACE_Atmoic_Op::value_i): Explicitly return - <value_> (by reference). This gives the user full, unrestricted - access to the underlying value. This method will usually be - used in conjunction with explicit access to the lock. Use with - care ;-) - -Sun Dec 14 21:37:08 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS.i (cond_wait,cond_timedwait): Fixed a missing return to - handle the case where we're using USYNC_PROCESS for the external - mutexes. - - * ace/OS: Made the use of the waiters_lock_ mutex consistent in - the ACE_cond_t implementation. This was broken since we used - the external_mutex to serialize access to the waiters_ count - inconsistently. Thanks to Patrick J. McNerthney - <pat@thememedia.com> for reporting this very subtle bug. - -Sun Dec 14 12:29:26 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * ace/Thread_Manager.cpp (exit,wait): Disabled thread join - functionality in Chorus. Thanks to Wei Chiang - <chiang@tele.nokia.fi> for providing the fix. - -Fri Dec 12 21:48:38 1997 Torbjorn Lindgren <tl@funcom.com> - - * ace/config-linux*.h: All Linux platform files except the - Linux-KCC modified to use the upgraded common Linux - configuration file. Linux-KCC looks reasonably easy to do, - if no-one that actually have KCC takes care of it I can do - this one later. - - * ace/config-linux-common.h: Modernized and updated the common - Linux configuration file to handle more of the common - configuration stuff. Also added the beginning of proper - glibc and egcs support for Linux. - -Sat Dec 13 14:22:28 1997 <irfan@TWOSTEP> - - * ace/Basic_Types.h: Changed _M_IX386 to _M_IX86. - -Fri Dec 12 19:41:42 1997 <irfan@TWOSTEP> - - * ace, tests, version_tests: Updated Win32 makefiles (VC4.2 and - VC5.0) to include new Basic_Types_Tests. - - * tests/Basic_Types_Test.cpp (main): Updated to use - ACE_OS::getpagesize(). - - * ace/ACE.cpp (round_to_pagesize): Updated to use - ACE_OS::getpagesize(). - - * ace/OS.i (getpagesize): Added this to ACE_OS. - -Fri Dec 12 07:49:01 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.i (inet_addr): changed type of "valid" from bool to u_int. - This code is only seen on VxWorks. - - * ace/IOStream.h (ACE_Quoted_String::operator <): changed return - type from bool to int. - - * ace/config-vxworks*.h,README: removed ACE_LACKS_SOCKET_BUFSIZ; - added ACE_DEFAULT_MAX_SOCKET_BUFSIZ of 32768. - - * examples/IPC_SAP/SOCK_SAP/C-in{client,server}.cpp: added - support to optionally set socket buffer sizes. - - * examples/System_V_IPC/SV_Semaphores/Semaphores_2.cpp: - removed unused constant SHMSZ. - - * examples/Timer_Queue/Reactor_Timer_Queue_Test.cpp: removed - unused constant NO_OF_IO_HANDLERS. - - * ace/IOStream_T.h (ipfx,ipfx1): don't call setstate () on WIN32, - because MS VC++ 5.0 doesn't declare it. Thanks to Irfan for - reporting this. This "fix" is temporary, maybe, until we - figure out the proper fix. - - * ace/OS.h: cast the 1 in ACE_MSB_MASK to an fd_mask, to support - 64-bit fd_masks. I suspect that with this change, the - ACE_HAS_BROKEN_BITSHIFT used in config-tandem.h is no longer - necessary. - - Added Basic_Types: - * ace/Basic_Types.*,Makefile: Basic_Types defines the - following preprocessor macros: - - Sizes of built-in types: - ACE_SIZEOF_CHAR - ACE_SIZEOF_WCHAR - ACE_SIZEOF_SHORT - ACE_SIZEOF_INT - ACE_SIZEOF_LONG - ACE_SIZEOF_LONG_LONG - ACE_SIZEOF_VOID_P - ACE_SIZEOF_FLOAT - ACE_SIZEOF_DOUBLE - ACE_SIZEOF_LONG_DOUBLE - - Wrappers for built-in types of specific sizes: - ACE_INT16 - ACE_UINT16 - ACE_INT32 - ACE_UINT32 - ACE_UINT64 - (Note: ACE_INT64 is not defined, because there is no ACE_LongLong for - platforms that don't have a native 8-byte integer type.) - - Byte-order (endian-ness) determination: - ACE_BYTE_ORDER, to either ACE_BIG_ENDIAN or ACE_LITTLE_ENDIAN - ACE_BIG_ENDIAN or ACE_LITTLE_ENDIAN - - * ace/OS.*: moved ACE_U_LongLong to Basic_Types.*. - - * ace/OS.h: added #include <float.h>. - - * ace/config-linux-common.h,README: added ACE_HAS_BYTESEX_H. - - * tests/Basic_Types_Test.cpp,run_tests.{bat,sh,vxworks),Makefile: - added new Basic_Types_Test. - - * tests/Time_Value_Test.cpp: removed ACE Version printout because - it is now in Basic_Types_Test. - -Fri Dec 12 03:00:21 1997 James C Hu <jxh@cs.wustl.edu> - - * ace/config-sunos5.5-egcs-1.0.h: Fundamemntally identical to g++ - version, except added ACE_HAS_STDCPP_STL_INCLUDES, which is - taken to mean that the Standard C++ STL include files are - supported. - - * IOStream.h: Added #if's to handle the case where - ACE_HAS_STRING_CLASS, but if ACE_HAS_STDCPP_STL_INCLUDES, then - the string class is found in <string>. Otherwise, it looks for - it in <String.h>. - - * ace/OS.i: Re-implemented strcasecmp methods. Thought bad - implementations were related to a JAWS bug, but the hand - implementations are not used on Solaris. - -Thu Dec 11 23:20:57 1997 <irfan@TWOSTEP> - - * ace/Strategies_T.cpp (activate_svc_handler): Changed the - parameter that was been passed into the open routine of - Svc_Handler. The <arg> that was passed to us is passed through - to open, rather than passing <this>. Thanks to Mats Nilsson - <mats.nilsson@xware.se> for pointing this out. - - * ace/Connector.cpp (activate_svc_handler): Made sure that we pass - the connector pointer as the void* parameter to the svc_handler. - -Thu Dec 11 23:35:04 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS.i. The code was always calling dlsum for the _init() - hook, even if dlopen() failed. In addition to fixing this, I - also removed a potential handle leak. Thanks to Patrick - J. McNerthney <pat@thememedia.com> for reporting this. - -Thu Dec 11 13:13:17 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/LSOCK.cpp (recv_handle): fixed to remove compile errors and - warnings on platforms with both ACE_HAS_4_4BSD_SENDMSG_RECVMSG - and ACE_HAS_THREADS. The delta from version 4.3 of this file - is very small. Thanks to Kirill Rybaltchenko - <Kirill.Rybaltchenko@cern.ch> for reporting this problem and - verifying the fix on LynxOS, the only affected platform. - - * ace/config-linux-kcc.h,include/makeinclude/platform_linux_kcc.GNU: - added linuxthreads support with the linux KAI C++ config. - Thanks to Ben Eng <ben@jetpen.com> for the patches. - - * ace/IOStream_T.h: iostream fx emulation now skips whitespaces and - flushes when tied. Thanks to Ben Eng <ben@jetpen.com> for the patch. - - * tests/IOStream_Test.cpp (server): check for double-quote instead - of newline as end-of-string character. Thanks to Ben Eng - <ben@jetpen.com> for the patch. - - * tests/SV_Shared_Memory_Test.cpp: renamed "allocator" to - "myallocator" to avoid conflict with STL. Thanks to Ben Eng - <ben@jetpen.com> for the patch. - - * include/makeinclude/platform_linux_lxpthread.GNU: protect - PRELIB definition with ifndef PRELIB, to allow overriding - without modifying the platform_macros.GNU file. Thanks to - James Johnson <jcej@lads.com> for this suggestion. - - * ace/IOStream_T.cpp: restored ACE_HAS_IOSTREAMH_INCLUSION - protection around the old g++ munch hack. There's no need - to trouble all (g++) platforms with its (4 byte) space penalty. - - * examples/IOStream/server/iostream_server.cpp (handle_input): - removed unused local "addr". - - * examples/Reactor/Dgram/{CODgram,Dgram}.cpp (run_test): added - return 0 at end so that the function always returns a value. - - * include/makeinclude/platform_lynxos.GNU: added comment indicator - to comment describing POSTLINK. - - * ACE-INSTALL.html: updated DEC CXX 6.0 test status: all tests pass - with the version of DIGITAL UNIX 4.0 and cxx 6.0 that we use. - -Wed Dec 10 23:57:00 1997 Chris Cleeland <cleeland@cs.wustl.edu> - - * ACE version 4.4.4, released Wed Dec 10 23:57:00 1997. - -Wed Dec 10 16:24:09 1997 <irfan@TWOSTEP> - - * ace/Auto_Ptr.i: Apparently, the standard states that in order to - access members of a templated superclass from a templated - subclass, then the member name must be fully qualified. - Therefore, changed access to p_ to ACE_Auto_Basic_Ptr<X>::p_. - Thanks to Stephen Coy <stevec@magna.com.au> for reporting this. - -Tue Dec 09 13:05:00 1997 <nw1@CHA-CHA> - - * ace/Containers.cpp (copy_all_nodes): Added a temporary - ACE_Node<T> pointer to get around the macro side effects. - -Sun Dec 7 10:20:46 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Service_Config.cpp (parse_args): Added an appropriate macro - to switch from char * to UNICODE. Thanks to Timothy A. Brown - <tabrown@montana.com> for reporting this. - - * ace/config-linux-*.h: Added ACE_HAS_GETRUSAGE_PROTO to all - the linux config files. Hopefully, this will help ACE to work - better with glibc. Thanks to Ernie Makris <emakris@ziplink.net> - and Torbjorn Lindgren <tl@funcom.no> for their help. - -Sat Dec 06 13:30:11 1997 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.3, released Sat Dec 06 13:30:11 1997. - -Sat Dec 6 12:02:09 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * examples/IOStream: Revised both the client and server to - demonstrate the ACE_IOStream functionality correctly. Also - added a README file explaining how to run the examples. Thanks - to James Johnson and Dustin Laurence - <laurence@alice.wonderland.caltech.edu> for suggesting this. - - * ace/IOStream_T: Added a new method eof() to the ACE_IOStream_T<> - class. This returns 1 if we've reached the end of the socket - stream, else 0. - -Fri Dec 5 16:58:29 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Addr.h: Added a const ACE_Addr & instead of const ACE_Addr. - This will avoid an extra copy. Thanks to Irfan for noticing - this. - - * examples/Reactor/Dgram/CODgram.cpp: Also updated the CODgram - test to be just like the Dgram test. - - * examples/Reactor/Dgram/Dgram.cpp (main): Fixed a bug where it - was sending a huge buffer instead of just the string size. - Also, exit once we've finished the send. Finally, revised the - test so that it is easier to understand and so that it exits - properly. Thanks to Michael Hoffman - <Hoffman_Michael@mac-mailserver.atc.ll.mit.edu> for reporting - this. - -Fri Dec 05 13:20:09 1997 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_vxworks5.x_ghs.GNU: added OFLAGS. - Thanks to Brian Mendel for suggesting these. - - * tests/IOStream_Test.cpp: updated to work with the standard C++ - library version of iostreams. That primarily involved not - using operator>>(char& c) and operator>>(char* c) for C strings. - Thanks to Ben Eng <ben@jetpen.com> for the patch. - - * ace/OS.{h,i},config-osf1-4.0.h: fixed sigwait on DEC CXX 6.0/ - recent DEC UNIX 4.0. The fix keys off the compiler version, - though it probably should depend on the OS. It adds a - sigwait declaration, which is a current signal.h deficiency. - With this change, ACE builds and tests cleanly on DEC UNIX. - - * ace/config-vxworks5.x-g++.h: restored comment that explains - ACE_HAS_MINIMUM_IOSTREAMH_INCLUSION. - - * netsvcs/lib/Makefile: the first line of the Local Target - (Server_Logging_Handler.*) override mysteriously disappeared, again. - -Fri Dec 5 12:15:49 1997 Chris Cleeland <cleeland@cs.wustl.edu> - - * ace/Select_Reactor.cpp: Made error messages on open failure more - verbose. Having these in place straightaway would have helped - track down some problems at Boeing much more effectively. - -Fri Dec 5 11:15:17 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * ace/Log_Msg.h (callers.): Added documentation for %I and %D - flags. Thanks to John Morey <jmorey@tbi.com> for reosting - this. - - * ace/Dynamic.h: - * ace/CORBA_Ref.h: - * ace/Token_Invariants.h: - * ace/Token_Collection.h: - * ace/Local_Tokens.h: - * ace/Remote_Name_Space.h: - * ace/Registry_Name_Space.h: - * ace/Strategies.h: - * ace/Name_Space.h: - * ace/Name_Request_Reply.h: - * ace/Naming_Context.h: - * ace/Name_Proxy.h: - * ace/Local_Name_Space.h: - * ace/Local_Name_Space_T.h: Corrected erroneous comments. These - files all contain incorrect filename info in the preambles which - make erroneous auto-generated man pages. - -Fri Dec 05 00:00:53 1997 <irfan@TWOSTEP> - - * ace/OS.h: If the user wants minimum IOStream inclusion, i.e., - they have defined ACE_HAS_MINIMUM_IOSTREAMH_INCLUSION, we will - just include the forward declarations, i.e., ace/iosfwd.h. Else - they will get all the stream header files, i.e., ace/streams.h. - The positive thing about this change is that if your code was - using stream classes and you were depending on OS.h to provide - you the stream header files, you will not have to change your - code, unless you have defined - ACE_HAS_MINIMUM_IOSTREAMH_INCLUSION. Thanks to Dr. Schmidt for - suggesting this and thanks to David Levine for suggesting it to - Dr. Schmidt ;-). - - * ace/config-vxworks5.x-g++.h: Restored - ACE_HAS_MINIMUM_IOSTREAMH_INCLUSION. - - * ace/Asynch_IO.cpp (open): Fixed typo. - -Thu Dec 04 21:20:47 1997 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.2, released Thu Dec 04 21:20:47 1997. - -Thu Dec 4 19:45:04 1997 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * ace/Object_Manager.cpp (ACE_Object_Manager_Destroyer): Fixed a - bug in the - ACE_Object_Manager_Destroyer::~ACE_Object_Manager_Destroyer - thread comparison routine. Thanks to Steven Coy for this. Also - renamed _saved_main_thread_id to saved_main_thread_id_, which is - proper ACE style. - - * Updated the ACE-INSTALL.html file to explain why you don't need - to build everything all the time. Thanks to Dustin Laurence - <laurence@alice.wonderland.caltech.edu> for suggesting this. - - * tests/SOCK_Connector_Test.cpp (fail_no_listener_nonblocking): - Added a check for ENETUNREACH. Thanks to Dustin Laurence - <laurence@alice.wonderland.caltech.edu> for reporting this. - - * ace/Asynch_Acceptor.cpp (open): Changed the local variable named - sap_any to sa to avoid conflicts with the new sap_any macro. - - * ace/Asynch_IO: Made the ACE_Service_Hanlder::open() method - non-virtual to work around bugs with Borland C++. Thanks to - Valik Solorzano Barboza <valik@geodan.nl> for suggesting this. - - * ace/ACE.cpp (COMPUTE): Put parens around elements the COMPUTE - macros so that it will work on Borland C++. Thanks to Valik - Solorzano Barboza <valik@geodan.nl> for suggesting this. - - * ace/config-win32-common.h: Added ACE_HAS_BROKEN_SAP_ANY to deal - with Borland C++ bugs. Thanks to Valik Solorzano Barboza - <valik@geodan.nl> for suggesting this. - - * ace/Addr: Added a workaround for the Borland C++ problems. This - introduces a new global macro called sap_any. However, this is - only used for Borland, so it should be ok. Thanks to Valik - Solorzano Barboza <valik@geodan.nl> for suggesting this. - - * ace/LSOCK.cpp (recv_handle): Make the code smarter if we've got - both STREAMS and 4_4BSD_SENDMSG_RECVMSG. - - * Reran "make depend" over all of ACE. - - * netsvcs/clients/Logger/indirect_logging.cpp (main): Changed - LOGGING to LOGGER. Thanks to Irfan for noticing this. - - * netsvcs/lib/Client_Logging_Handler: Changed "rendezvous" to - "logger." - - * netsvcs/clients/Logger/indirect_logging.cpp, - netsvcs/lib/Client_Logging_Handler: Changed DEFAULT_RENDEZVOUS - to DEFAULT_LOGGER_KEY. Thanks to Steven Coy for pointing out - this inconsistency. - - * ace/OS.h: Renamed ACE_LOGGER_KEY to ACE_DEFAULT_LOGGER_KEY to - make it consistent with other ACE_DEFAULT_* names. - - * ace/OS.h: Changed the definition for ACE_LOGGER_KEY so that it - defaults to "localhost:10015" on platforms that don't support - ACE STREAM Pipes. Thanks to Steven Coy for suggesting this. - - * ace/Service_Config.cpp: Added the -k option to parse_args() so - that the user can set the logger key. Thanks to Steven Coy for - suggesting this. - - * ace/Service_Config: Added the logger_key as a parameter to the - other open() method, as well. Thanks to Steve Coy for being - picky ;-). - - * netsvcs/clients/Logger/direct_logging.cpp (main): Modified the - code to use the new framing scheme. - - * netsvcs/lib/Server_Logging_Handler_T.cpp - (handle_logging_record): Updated the code to take advantage of - our new framing scheme. - - * netsvcs/lib/Client_Logging_Handler.cpp (send): We can be smarter - now about sending the ACE_Log_Record across the network since - the length field is now fixed and it's guaranteed to be in the - first 4 bytes. Thanks to Steve Coy for suggesting this. - - * ace/Log_Record.h (ACE_Log_Record): Replaced the horribly - non-portable "long length_" field with the "ACE_INT32 length_" - field, which is nice because it is portable across platforms - with different long-word sizes. - -Thu Dec 04 09:57:40 1997 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_osf1_4.GNU: allow enabling of - WARNING_FLAGS by setting CXX_VER to POST_5X. Thanks to - James Johnson for this suggestion. - - * ace/Svc_Conf_l.cpp: rebuilt with Doug's flex. - -Thu Dec 04 03:46:04 1997 <irfan@TWOSTEP> - - * ace/Makefile (Svc_Conf_l.cpp): Changed the sed macros to produce - OS.h include instead of stdcpp.h. - - * ace/Svc_Conf_l.cpp: Updated after code was regenerated using the - new sed changes in the Makefiles. Currently there are two - inclusions of OS.h in this file and an extra stdlib.h. Maybe - someone more experienced in sed can take these out by updating - the Makefile. - - * ace/IOStream_T.h (stdcpp.h): Removed inclusion of stdcpp.h. It - is redundant because IOStream.h includes it. - - * ace/streams.h: New file. This files tries to reduce the - differences between the old and new stream libraries for the - user. This is done by automatically selecting the correct - includes depending on the ACE_HAS_STANDARD_CPP_LIBRARY variable. - This file also helps out by making commonly used classes (e.g., - ios, ostream, iostream, cin, cout, etc) available in the global - name space when used with the Standard C++ Library. Use this - file where you would usually include <iostream.h> or <iostream>. - - * ace/iosfwd.h: New file. This files deals with forward - declaration for the stream classes. Remember that since the new - Standard C++ Library uses templates for the implementation of - the streams, simple forward declaration will not work. Classes - forward declared by this class include: - - ios - streambuf - istream - ostream - iostream - filebuf - ifstream - ofstream - fstream - - Use this file where you would usually forward declare the - classes listed above. - - * ace/stdcpp.h: Removed this file from the repository. This file - was trying to do too much. All the inclusion of the C Library - headers were unnecessary. This is because the new C++ standard - preserves the old C library include files. The functions from - those files will still be in the global namescape. The new - header files (e.g., cstdio), provides the same prototypes as the - old files (e.g., stdio.h), except that all prototypes are in the - std namespace. Since we want the old behavior and global - prototypes (for backward compatibility), inclusion of the new - header files was incorrect. - - stdcpp.h was included by OS.h. This caused almost all files to - force the inclusion of the stream header files. This was - changed so that OS.h only includes the ace/iosfwd.h file. This - means that if your code was using stream classes and you were - depending on OS.h to provide you the stream header files, you - will have to change your code to either include iostream.h or - iostream, or more portably ace/streams.h. - - * ace/OS.h: Removed stdcpp.h. Moved the standard C library - includes from stdcpp.h to OS.h. Also added ace/iosfwd.h to give - forward declaration for the stream classes. - - * ACE: Added explicit inclusion of ace/streams.h to the following - files: - - ace/IOStream.cpp - ace/Log_Record.cpp - ace/OS.cpp - ace/SString.cpp - ace/Log_Msg.cpp - tests/test_config.h - examples/Log_Msg/test_log_msg.cpp - examples/Reactor/Misc/test_time_value.cpp - examples/Threads/task_three.cpp - netsvcs/lib/Logging_Strategy.cpp - examples/Logger/Acceptor-server/server_loggerd.cpp - examples/ASX/UPIPE_Event_Server/event_server.cpp - examples/Reactor/Ntalker/ntalker.cpp - examples/Shared_Malloc/test_persistence.cpp - examples/System_V_IPC/SV_Message_Queues/MQ_Server.cpp - examples/System_V_IPC/SV_Message_Queues/MQ_Client.cpp - examples/System_V_IPC/SV_Message_Queues/TMQ_Server.cpp - examples/System_V_IPC/SV_Message_Queues/TMQ_Client.cpp - - * Updated makefile dependencies throughout ACE to reflect the - removal of stdcpp.h and the addition of streams.h. - - * examples/Threads/barrier2.cpp: Removed inclusion of stdcpp.h. - - * ace/IOStream_T.cpp: Reorganized include files so that inclusion - of config.h is not necessary. - - * ace/IOStream.cpp: Changed inclusion of config.h to OS.h. - - * ace: Removed config.h from the following files. It was - redundant. - - Timer_Hash.cpp - Timer_Heap.cpp - Timer_List.cpp - Timer_Queue.cpp - Timer_Wheel.cpp - - * ace: Added code to protect against multiple inclusions of - config.h. Following files were effected: - - Trace.cpp - Log_Msg.cpp - - * ace/config-win32-common.h: Removed redundant definitions of - ACE_HAS_TYPENAME_KEYWORD and ACE_HAS_SIG_ATOMIC_T. Thanks to - Matthias Kerkhoff <make@cs.tu-berlin.de> for pointing this out. - -Wed Dec 3 21:37:47 1997 Douglas C. Schmidt <schmidt@flamenco.cs.wustl.edu> - - * Added the new ACE-inheritance.ps and pdf files, which show the - class structure of ACE. Thanks to John Lindal - <jafl@cheshire-cat.caltech.edu> for generating this. - - * ace/Service_Config: Removed the static logger_key_ member and - replaced it with a (defaulted) parameter to open(). This is - more flexible. Thanks to Steve Coy for the suggestion. - - * ace/Service_Config.h: Addressed all the "@@" questions. - - * examples/IPC_SAP/SOCK_SAP/CPP-inclient.cpp (main): Fixed a - subtle bug in the "quit" logic. Thanks to Michael Hoffman - <Hoffman_Michael@mac-mailserver.atc.ll.mit.edu> for reporting - this. - - * netsvcs/lib/Server_Logging_Handler_T.cpp (handle_logging_record), - netsvcs/lib/Client_Logging_Handler.cpp (send): - Changed the code so that we now always send ACE_UINT32 bytes - worth of data for the length field. This ensures we're portable - across platforms. Thanks to Steven Coy for noticing this. - -Wed Dec 03 18:02:09 1997 <irfan@TWOSTEP> - - * performance-tests/Synch-Benchmarks/Benchmark.cpp: Added template - instantiations for Atomic_Op. - -Wed Dec 3 10:29:58 1997 Carlos O'Ryan <coryan@macarena.cs.wustl.edu> - - * ace/ACE.cpp: - Use ACE_UINT32 to compute the 32-bit CRC, on Alphas (and - probably other 64 bit machines) u_long seems to be longer than - that. - -Wed Dec 03 07:39:15 1997 Steve Huston <shuston@riverace.com> - - * include/makeinclude/platform_hpux.GNU: - * include/makeinclude/platform_hpux_gcc.GNU: Moved some compile options - from the DCFLAGS to the CCFLAGS to compile correctly with or without - debugging. - -Wed Dec 03 06:26:14 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/Future.cpp, - tests/{Future_Test,Reactors_Test,Reader_Writer_Test.cpp, - examples/Threads/manual_event.cpp: - moved ACE_Atomic_Op<ACE_Thread_Mutex, int> instantiation from - ace/Future.cpp to where it is used. - - * ace/OS.cpp (inet_aton): return 0 if addr argument is 0. Thanks - to Wei for reporting this. - - * ace/config-linux.common.h: set an ACE_DEFAULT_BASE_ADDR, on Alpha - for now. This might be necessary on Intel as well. - - * ace/OS.{h,i}, config-osf1-4.0.h: fixed use of ::sigwait with - DIGITAL_UNIX's cxx 6.0, which now uses ACE_HAS_ONEARG_SIGWAIT. - Also, added an ACE_DEFAULT_BASE_ADDR to the config file. - - * tests/{Priority_Reactor_Test,Reactor_Performance_Test}.cpp: - instantiate ACE_Map_*, for use by ACE_Connector, with ACE_HANDLE - instead of int. That's the way it's declared in ace/Connector.h. - -Wed Dec 03 00:17:50 1997 <irfan@TWOSTEP> - - * tests/test_config.h: Removed code for - ACE_HAS_MINIMUM_IOSTREAMH_INCLUSION, as this option is no longer - necessary. - - * ace/stdcpp.h: Removed code for - ACE_HAS_MINIMUM_IOSTREAMH_INCLUSION, as this option is no longer - necessary. - - * ace/config-vxworks5.x-g++.h: Removed - ACE_HAS_MINIMUM_IOSTREAMH_INCLUSION. This flag is no longer - required. - - * ace/IOStream_T.cpp: Removed ACE_HAS_MINIMUM_IOSTREAMH_INCLUSION - from #if defined. - - * ACE: Removed the explicit iostream.h inclusion from these: - - examples/ASX/UPIPE_Event_Server/event_server.cpp - examples/Log_Msg/test_log_msg.cpp - examples/Logger/Acceptor-server/server_loggerd.cpp - examples/Reactor/Misc/test_time_value.cpp - examples/Shared_Malloc/test_persistence.cpp - examples/Threads/task_three.cpp - netsvcs/lib/Logging_Strategy.cpp - - * ace/IOStream.h: Darrell missed removing this explicit iostream.h - inclusion. - - * ace/Future.cpp: Removed template instantiation of Atomic_Op - since Atomic_Op is not used by Future. - -Fri Dec 02 11:46:17 1997 <nw1@CHA-CHA> - - * ace/Proactor.cpp (ACE_Proactor): Reenabled Proactor's thread - manager to wait on its Timer_Handler thread. - - * ace/Thread_Manager.cpp (close_singleton): Removed Win32 specific - codes. This is no longer necessary with above changes. - (wait): Close down thread handles when removing detached - threads. Avoid performing wait when the program is shutting - down. - (remove_thr_all): Extracted out this protected member function. - (close): Changed to use the new remove_thr_all function. - - * ace/Proactor.cpp (handle_events): Changed the magic number 258 - to WAIT_TIMEOUT. Thanks to Mark Rabotnikov - <mark@netmanage.co.il> for pointing this out. - -Tue Dec 02 14:13:15 1997 Steve Huston <shuston@riverace.com> - - * include/makeinclude/platform_hpux_aCC.GNU: Moved some compile options - from the DCFLAGS to the CCFLAGS to compile correctly with or without - debugging. Thanks to Istvan Buki <ibuki@fedex.com> for reporting - the problem. - -Tue Dec 02 06:16:39 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/Object_Manager.cpp (~ACE_Object_Manager_Destroyer): don't - delete the ACE_Object_Manager if not destroyed by the main - thread. That can cause problems, at least on WIN32. And - it should only happen under exceptional conditions, at least - on WIN32. Thanks to Ari Erev <ari_erev@icomverse.com> and - John Neystadt <john_neystadt@icomverse.com> for suggesting this. - - * ace/IOStream.cpp (underflow,overflow): moved get_mode_ and - put_mode_ accesses so that it now works on KAI C++. - Thanks to Ben Eng <ben@jetpen.com> for the patch. - - * ace/config-linux-kcc.h: commented out ACE_TEMPLATES_REQUIRE_SOURCE. - Thanks to Ben Eng <ben@jetpen.com> for this update. - - * ace/Profile_Timer.cpp (elapsed_time): changed - ACE_ONE_SECOND_IN_USECS to ACE_ONE_SECOND_IN_NSECS. Thanks - to Wei for noticing this. - - * ace/OS.cpp (inet_aton): removed CHORUS specialization. - Thanks to Wei for reporting this. - - * include/makeinclude/platform_chorus.GNU: added CHORUS = 1, - for use by individual Makefiles. - - * examples/Shared_Malloc/test_persistence.cpp: removed - ACE_Malloc<ACE_MMAP_MEMORY_POOL, ACE_Null_Mutex> - instantiation because it's in ace/System_Time.cpp. - - * ace/config-lynxos.h: removed ACE_HAS_UALARM, because LynxOS - apparently doesn't. - - * include/makeinclude/platform_lynxos.GNU: added -Winline and - removed -fno-strict-prototypes. - - * tests/Message_Queue_Test.cpp (main): increased queue size from - default of 16Kb to 32Kb, so that it runs on machines with 8Kb - pagesizes. And, added a check to see if the queue is full - before enqueuing. - - * ace/OS.i, config-osf1-4.0.h: DECXX 6009006 has SIGWAIT. - - * ACE-INSTALL.html: updated DEC_UNIX test status. Message_Queue_Test - and all of the Mutex_Tests now pass. Conn_Test no longer does. - -Tue Dec 02 00:18:04 1997 <irfan@TWOSTEP> - - * ace/Memory_Pool.cpp (handle_signal): If guess_on_fault_ is true, - then we cannot call remap(0). This will always put the address - out of range. Therefore, we will shoot in the dark (user has - instructed us to do so as guess_on_fault_ is true) and try to - remap the file. If the user does not want this behavior, the - guess_on_fault_ flag should be turned off. Thanks to Jon - Prettyman <jonp@als.ameritech.com> for pointing this out. - -Mon Dec 01 16:13:06 1997 <irfan@TWOSTEP> - - * ace/config-win32-common.h: Added back ACE_LACKS_ACE_IOSTREAM to - the config file. The test is not compiling. - - * tests/run_tests.bat: Look for "Win32 structured exception" in - the log files. - -Mon Dec 1 14:01:02 1997 Darrell Brunsch <brunsch@cs.wustl.edu> - - * ace/Log_Msg.cpp, ace/Log_Msg.h, ace/Log_Record.cpp, - ace/Log_Record.h ace/OS.h, ace/stdcpp.h, tests/test_config.h: - - Changed use of ACE_HAS_MINIMUM_IOSTREAMH_INCLUSION in stdcpp.h - to include iostream when the above is #defined. Made - changes to other files to accommodate this change. - -Mon Dec 01 12:15:00 1997 <irfan@TWOSTEP> - - * ace/config-win32-common.h (ACE_LACKS_ACE_IOSTREAM): James CE - Johnson claims that ACE_LACKS_ACE_IOSTREAM is no longer not - necessary when using the ACE_HAS_STANDARD_CPP_LIBRARY. - - * ace/OS.h: Changed the definition of ACE_UNIMPLEMENTED_FUNC so - that it just evaporates on platform that have - ACE_REQUIRES_FUNC_DEFINITIONS defined. - - ACE_UNIMPLEMENTED_FUNC is only required for template classes and - not for a normal class. When the template is explicitly - instantiated, it causes some compilers to look for the - definition of all the declared methods. - - On platforms that have this behavior, we have decided to take - out the declaration of this function. This is necessary since - an empty definition of {} usually does not work with some - constructors. And by providing a explicit definition of this - function, will cause all platforms to suffer. - - Therefore, platforms with ACE_REQUIRES_FUNC_DEFINITIONS defined - must be specially careful when using this. - -Mon Dec 01 07:56:02 1997 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_sunos5_sunc++.GNU: added - -xildoff to LDFLAGS, to avoid occasional problems with - the incremental linker. Thanks to Hans Rohnert - <Hans.Rohnert@mchp.siemens.de> for this suggestion. - - * include/makeinclude/platform_vxworks5.x_g++.GNU, - etc/ace_ld, bin/ace_ld, ACE-INSTALL.html: - moved ace_ld from etc/ to bin/. - - * Makefile: added etc/ to CLONE and RELEASE_FILES. - - * include/makeinclude/platform_lynxos.GNU: added -Wall to CFLAGS. - - * ace/ACE.cpp (set_handle_limit): only call setrlimit if - RLIMIT_NOFILE is #defined. - - * ace/OS.h: removed ACE_HAS_STDARG_THR_DEST dependence for - ACE_THR_DEST definition. - - * ace/OS.cpp (thr_keycreate): with ACE_HAS_DCETHREADS only, cast - dest argument to ::pthread_keycreate if ACE_HAS_STDARG_THR_DEST. - - * ace/OS.i (get{host,proto,serv}by{addr,name}_r): with - ACE_HAS_NONCONST_GETBY only, added some ACE_UNUSED_ARGS. - - * examples/Shared_Malloc/Malloc.cpp: removed - ACE_Guard<ACE_Process_Mutex> instantiation because it's in - ace/Synch.cpp. - - * ace/config-lynxos.h: added ACE_HAS_RECURSIVE_THR_EXIT_SEMANTICS. - Without it, the Thread_Manager went recursively ballistic. - - * performance-tests/Misc/preempt.cpp (main): fixed printout if - ACE_OS::sched_params () call failed. - -Sun Nov 30 22:44:50 1997 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.4.1, released Sun Nov 30 22:44:50 1997. - -Sun Nov 30 20:32:50 1997 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_osf1_3.2.GNU: removed -ptr option - from CCFLAGS, so the default template repository will be used. - Thanks to Dani Flexer <danif@ivory-sw.com> for reporting this. - - * ace/Svc_Conf_l.cpp,Makefile: - 1) #include ace/stdcpp.h instead of stdio.h so that anything - that needs to be #included with/before stdio.h can be. - 2) Changed ECHO to ACE_SVC_CONF_ECHO to avoid compiler warning if - ECHO was #defined to be a constant. - - * ace/Makefile: set ACELIB to null so that it can be used to build - shared libs, including libACE.$(SOEXT). - - * include/makeinclude/platform_osf1_4.0.GNU: added $(ACELIB) to - SOFLAGS to avoid shared lib build warnings about undefined ACE - symbols. - - * ace/Thread_Manager.h: declare ACE_THR_MEMBER_FUNC outside of - ACE_Thread_Manager class with __GNUG__ instead of with - VXWORKS || IRIX5. The problem is with older versions of g++, - not with the OS. - -Sun Nov 30 14:13:09 1997 Douglas C. Schmidt <schmidt@merengue.cs.wustl.edu> - - * netsvcs/servers/README: Clarified a bit more about how to - configure the client and server logging processes. Thanks to - Neil Cohen for motivating this. - -Sun Nov 30 17:36:35 1997 Sergio Flores <sergio@tango.cs.wustl.edu> - - * ace/Synch.cpp (ACE_File_Lock): Fixed order of parameters in the - call to this->open() and too many arguments in call to - ACE_File_Lock::open(). - -Sun Nov 30 16:31:41 1997 Carlos O'Ryan <coryan@macarena.cs.wustl.edu> - - * ace/Synch.h: - * ace/Synch.cpp: - ACE_File_Lock should create the file if it does not exist, to do - so the constructor takes the <flags> parameter to ACE_OS::open() - and defaults to O_CREAT|O_RDWR. - -Sat Nov 29 21:31:07 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * ace/Proactor.cpp (~ACE_Proactor_Timer_Handler): Removed explicit - call to thr_mgr_.close(). - (ACE_Proactor): Be sure to ask Thread_Manager not to wait for - the timer_handler_ thread. - - * ace/Thread_Manager.cpp (close_singleton): Only reset - exit_on_wait status if we are responsible for deleting the - global thread manager. - -Sat Nov 29 15:15:04 1997 Carlos O'Ryan <coryan@swarm.cs.wustl.edu> - - * ace/IPC_SAP.cpp: - Removed warning for functions not returning a value on SGI, it - seems that the flow changed and the last return was reached - sometimes, hence the ACE_NOTREACHED macro is no longer needed. - -Sat Nov 29 13:12:47 1997 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Synch: Conditionally inlined all the ACE_Process_Mutex - wrapper calls and moved them into the Synch.i file. I'm not - sure why there weren't here before. - - * ace/Synch: Reimplemented ACE_RW_Process_Mutex to use - ACE_File_Lock, which provides the proper readers/writer - semantics that had been lacking in the previous implementation. - -Fri Nov 28 04:25:23 1997 Nanbor Wang <nw1@merengue.cs.wustl.edu> - - * ace: Removed redundant trailing commas from - ACE_UNIMPLEMENTED_FUNC. - - * ace/Thread_Manager.{h,i,cpp}: Delete declaration of - remove_thr_self. It is no longer in use. - (wait_on_exit): Added this function to let users specify whether - they want their Thread_Manager wait for spawned threads on exit - or not. - (wait): On Win32, handles should be closed if we decide to let - the thread go wild. - (remove_thr): Added another parameter to control whether - remove_thr should close the handle or not. It is only used on - Win32. - (close): Removed the <automatic_wait> parameter for it has been - incoporated as a member. - (close_singleton): Reset <automatic_wait_> on Win32. We'll never - need to wait for unfinished threads on NT. - (ACE_Thread_Manager): Added initialization of member - <automatic_wait_>. It is default to 1 (wait on exit.) - - * ace/Containers.h: Corrected erroneous comments. - -Wed Nov 26 10:33:22 1997 <irfan@TWOSTEP> - - * ace/Log_Msg (msg): Changed the parameters from char * to const - char *. - - * ace: Added unimplemented copy constructors and assignment - operators to the following classes: - - ACE_Map_Manager - ACE_Message_Block - ACE_Data_Block - ACE_Log_Msg - ACE_Message_Queue - ACE_Task - ACE_Task_Base - - Added ACE_UNIMPLEMENTED_FUNC macros to these classes: - - ACE_Mem_Map - ACE_Priority_Reactor - ACE_Select_Reactor - ACE_WFMO_Reactor - ACE_XtReactor - ACE_Reactor - ACE_Read_Buffer - ACE_CORBA_Handler - ACE_TSS_TPQ_Entry - ACE_Object_Manager - ACE_File_Lock - ACE_Semaphore - ACE_RW_Mutex - ACE_Mutex - ACE_Null_Barrier - ACE_Null_Condition - ACE_Null_Mutex_Guard - ACE_Event - ACE_Thread_Mutex - ACE_Thread_Mutex_Guard - ACE_Condition_Thread_Mutex - ACE_Recursive_Thread_Mutex - ACE_Barrier - ACE_Condition - - Thanks to Stanley D Leeson <STANLEY.D.LEESON@cdev.com> for these - suggestions. - -Sat Dec 20 14:57:30 1997 <nw1@CHA-CHA> - - * All MSVC 5.0 project files: Removed unnecessary library - inclusions. Most of them are not necessary. Thanks to - Patrick J. McNerthney <pat@thememedia.com> for suggesting - this. - -Wed Nov 26 15:49:14 1997 David L. Levine <levine@cs.wustl.edu> - - * ace/IOStream_T.i (recv) affects WIN32 only: fixed typo, - ACE_BIT_ENABLED instead of ACE_BIT_ENTABLED IOStream_T.i. - Thanks to James CE Johnson <jcej@lads.com> for reporting this. - - * include/makeinclude/platform_lynxos.GNU: added information - on necessary OS patches. - -Wed Nov 26 09:45:43 1997 Chris Cleeland <cleeland@cs.wustl.edu> - - * ace/OS.h (ACE_PROPER_SINGLETON_INSTANTIATION): Corrected - misspelling in the "! defined (MSC_VER)" branch. I just happened - to notice this, so I don't know if anybody ever tried it and it - failed. - -Tue Nov 25 12:07:07 1997 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_osf1_3.2.GNU: updated for - OSF/1 3.2g/CXX 5.7. Thanks to Dani Flexer <danif@ivory-sw.com> - for the updates. - -Tue Nov 25 01:07:39 1997 <irfan@TWOSTEP> - - * examples/Reactor/Proactor/test_multiple_loops.cpp (main): Made - sure that the Proactor was removed from reactor. - - * ace/WFMO_Reactor.cpp (remove_handler_i, - remove_suspended_handler_i): Make sure to preserve DONT_CALL - with normal event entries. - - * ace/Proactor.cpp: - - - A flag used to indicate when we are shutting down. - - - Made sure to signal the timer event when we are closing down. - - - Thread Manager: don't bother to wait (since the thread may have - already gone). But make sure to close up the descriptor. This - may not be necessary in the future when THR_DETACHED is - correctly implemented. - - * ace/Asynch_IO.h: Changed some code from protected to public. - Even though these functions are really protected, sometimes it - simplifies user code to be able to "fake" a result. Users - should use these with care. - - * examples/Reactor/Proactor/test_proactor.cpp - (handle_write_stream): Added code for partial write to socket. - However it ever seems to do a short write ;-). Also made sure - that the initial data coming from the AcceptEx call is also - dumped to the file. Thanks to Mark Rabotnikov - (mark@netmanage.co.il) for pointing this out. |