path: root/ChangeLog-99b

Thu Jan 14 02:33:37 1999  Douglas C. Schmidt  <schmidt@cs.wustl.edu>

        * ACE version 4.6.13 released.

Thu Jan 14 00:07:34 1999  Irfan Pyarali  <irfan@cs.wustl.edu>

	* docs/tutorials/014/stream.cpp: Added #include "ace/streams.h".

Wed Jan 13 23:18:03 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

	* ace/OS.h:
        * config-g++-common.h:
        * config-osf1-4.0.h:
        * config-psos-diab.h:
        * config-psos-tm.h: Revert my previous change.  Removed the
	  ChangeLog entry about the change.

Wed Jan 13 22:30:49 1999  David L. Levine  <levine@cs.wustl.edu>

        * include/makeinclude/platform_chorus_ghs.GNU: removed
          explicit link with libedgnoe.a, because GreenHills adds
          that implicitly.

        * ace/OS.h: added quick hack to ACE_HAS_BROKEN_EXPLICIT_DESTRUCTOR
          logic to allow compilation with g++.

Wed Jan 13 18:48:30 EST 1999  James CE Johnson  <jcej@lads.com>

        * docs/tutorials/004/client.cpp:
          docs/tutorials/010/message_queue.cp:
          docs/tutorials/011/message_queue.cpp:
          docs/tutorials/012/message_queue.cpp:
          docs/tutorials/013/message_queue.cpp:
          docs/tutorials/014/EndTask.h:
          docs/tutorials/019/client.cpp:
          docs/tutorials/019/client2.cpp:
          docs/tutorials/019/server.cpp:
          docs/tutorials/019/server2.cpp:
          docs/tutorials/019/shmem.cpp:
          docs/tutorials/020/client.cpp:
          docs/tutorials/020/server.cpp:
          docs/tutorials/021/client.cpp:
          docs/tutorials/021/mpool.cpp:
          docs/tutorials/021/mpool.h:
          docs/tutorials/021/server.cpp:
                Fixed many NT issues found by Irfan.  Most are related to the
                fact that NT doesn't have SysV shared memory.

Wed Jan 13 04:45:59 1999  James C Hu  <jxh@cs.wustl.edu>

        * ace/Cache_Object.{h,cpp}:
        * ace/Cache_Manager.{h,cpp}:
        * ace/Cache_Manager_T.{h,cpp}:
        * ace/Cache_Hash_T.{h,cpp}:
        * ace/Cache_Heap_T.{h,cpp}:
        * ace/Hash_Bucket_T.{h,cpp}:
          Supporting infrastructure for the new Filecache.  These classes
          form the basis of a generic in memory cache engine.

Wed Jan 13 02:17:57 1999  Irfan Pyarali  <irfan@cs.wustl.edu>

        * ace/OS.cpp (writev): This function should use ACE::write_n
          instead of ACE::send_n.  Similarly, readv() should use
          ACE::read_n instead of ACE::recv_n.

        * ace/ACE.cpp (send): This function should use ACE_OS::sendv
          instead of ACE_OS::writev.  Similarly, recv() should use
          ACE_OS::recvv instead of ACE_OS::readv.

        * docs/tutorials:  Fixed all the NT project files.

Tue Jan 12 22:41:05 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

        * ace/config-freebsd.h: Uncommented ACE_HAS_SIGWAIT.  This is
          required for FreeBSD 2.2.8. Thanks to John Aughey
          <jha@FreeBSD.ORG> for reporting this.

        * ace/SOCK_IO.{h,i}:
        * ace/SOCK_Stream.{h,i} (sendv_n/recvv_n): Moved these functions
          from SOCK_IO to SOCK_Stream where it makes more sense to have
          the semantic of sending/receiving <n> bytes.

Tue Jan 12 19:52:26 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

        * ACE-INSTALL.html: Added more info on dynamically linking
          run-time libraries.

        * ace/SOCK_IO.{h,i,cpp} (sendv,recvv,sendv_n,recvv_n): Renamed
          vector send/recv.  There were name clashing since these methods
          have similar signatures to some send_n/recv_n in
          ACE_SOCK_Stream, they got hidden by the derived methods.
          Renaming them solve the problem.

Tue Jan 12 10:59:58 1999  David L. Levine  <levine@cs.wustl.edu>

        * include/makeinclude/platform_chorus_ghs.GNU: fixed location of
          libedgenoe.a, so that a symlink in $(GHS_DIR) is no longer necessary.
          Thanks to Steve Kay for reporting this.

        * ace/config-mklinux.h: removed ACE_HAS_SOCKLEN_T, addedUSE_XOPEN.
          Thanks to Andreas Tobler <toa@pop.agri.ch> for these updates.

Tue Jan 12 08:44:11 EST 1999  Aniruddha Gokhale  <gokhale@sahyadri.research.bell-labs.com>

        * ace/Acceptor.cpp (handle_close):

          Since the reactor_ data member is made private, we need to use
          its accessor method to retrieve it.

Tue Jan 12 02:12:38 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

        * ace/OS.h (ACE_DES_NO_FREE,ACE_DES_FREE): These two macros
          shouldn't be treated differently no matter
          ACE_HAS_BROKEN_EXPLICIT_DESTRUCTOR is defined or not.
          Otherwise, virtual destructor won't work in one of the original
          definitions.

          There shouldn't be different different definitions for calling
          template destructor explicitly either.  However, we need to find
          that out.  The meaning of ACE_HAS_BROKEN_EXPLICIT_DESTRUCTOR
          seems to be reversed.  That needs to be fixed also.

Tue Jan 12 00:18:15 1999  Douglas C. Schmidt  <schmidt@ace.cs.wustl.edu>

        * ace/Event_Handler.h: Moved the priority_ and reactor_ data
          members into the private section of the class since these can
       always be accessed via their accessor methods.  Thanks to Mike
          Goldman <whig@by.net> for reporting this.

Mon Jan 11 15:30:26 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

        * ace/Process.{h,i} (get_process_attributes,get_thread_attributes):
          Removed the constness from the return value to avoid Intel C++
          warnings.  Thanks to Karel Zuiderveld
          <kzuiderveld@vitalimages.com> for reporting this.

        * bin/ADDIDL.DSM: A VB script to add new IDL files into DevStudio
          projects.   Thanks to Peter <weatp@syntron.com> for contributing
          this nice tool.

Mon Jan 11 12:54:26 1999  Douglas C. Schmidt  <schmidt@tango.cs.wustl.edu>

        * ace/SOCK_IO: Added const qualifiers to all the send*() and
          recv*() methods that take ACE_Time_Value values.  Thanks to Jody
          Hagins <jody@atdesk.com> for reporting this.

        * ace/SOCK_IO: Added a new send_n() method that uses the new
          ACE::sendv_n() method!

        * ace/ACE: Added a new sendv_n() and writev_n() method that sends
          all the bytes in the iovec!

        * ace/ACE.h (ACE): Added default values of 0 for ACE::writev() and
          ACE::readv().

        * examples/NT_Service: Tidied up the formatting to conform to the
          ACE programming guidelines.

Mon Jan 11 10:23:06 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

        * ace/config-freebsd-pthread.h: Added ACE_LACKS_READDIR_R.
        * ace/config-freebsd.h: Added ACE_LACKS_READDIR_R,
          ACE_HAS_NONCONST_MSGSND, ACE_LACKS_MALLOC_H.  ACE_HAS_SIGINFO_T
          only applies to 3.0 and above.  Thanks to Ivan Pascal
          <pascal@info.tsu.ru> for reporting the change.

Sun Jan 10 21:52:57 1999  David L. Levine  <levine@cs.wustl.edu>

        * tests/Message_Queue_Notifications_Test.cpp (iterator_test):
          added comment explaining why a message queue size of 32 Kb
          is used, instead of the default of 16 Kb.

        * docs/ACE-guidelines.html: added guidelines for boolean types
          and function return values.

Sun Jan 10 17:33:29 1999  Douglas C. Schmidt  <schmidt@ace.cs.wustl.edu>

        * tests/Message_Queue_Notifications_Test.cpp: Replaced delete mb
          with mb->release().  Thanks to Susan Liebeskind
          <susan.liebeskind@gtri.gatech.edu> for reporting this
          inconsistency.

        * ace/Task_T.h,
          ace/Message_Queue_T.h: Clarified the fact that the
          ACE_Time_Value arguments to the Message_Queue methods
          use absolute, rather than relative, time.  Thanks to
          Stanford S. Guillory <sguillory@vignette.com> for pointing out
          the need for this clarification.

Sun Jan 10 09:18:41 1999  Martin Krumpolec <krumpo@pobox.sk>

        * ace/Log_Msg.cpp (log): only re-enable tracing if it had
          not been explicitly disabled.

Sun Jan 10 01:06:18 1999  Irfan Pyarali  <irfan@cs.wustl.edu>

        * ace/Log_Msg.cpp (log): Change call from exit() to abort().

        * ace/OS.i (abort): Added new function.

Sat Jan  9 22:05:07 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

        * bin/tao_env.pl: A new perl script that help you translate _env
          to TAO_IN_ENV.

Sat Jan 09 10:53:37 1999  David L. Levine  <levine@cs.wustl.edu>

        * ace/Name_Proxy.cpp (open): swapped branches of conditional
          so that options.time_value () is used if USE_TIMEOUT is enabled.
          Thanks to Mike Goldman <whig@by.net> for reporting this.

Fri Jan 08 19:19:17 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

        * ace/OS.h: Applied ACE_TEXT to the definition of
          ACE_DEFAULT_TEMP_FILE on Win32.  Thanks to
          <gwross@west.raytheon.com> for reporting this.

Fri Jan 08 13:24:12 1999  David L. Levine  <levine@cs.wustl.edu>

        * ace/Synch.h (ACE_Process_Mutex): made data members
          private instead of public.  Thanks to Peter Gross
          <pgross@signalsoftcorp.com> for reporting this.

        * examples/Threads/reader_writer.cpp: removed volatile
          qualifier from declaration of "shared_thr_id", so that
          the file will compile on DU 4.0 with DCE threads.  The
          volatile qualifier caused a type mismatch with
          ACE_thread_t.  And, it's not necessary, because all
          access of the shared_thr_id is guarded.

        * include/makeinclude/platform_chorus_ghs.GNU: added gnuch68
          to -alttools, and removed bin from AR; use libedgnoe.a instead
          of libedg.a with GHS 1.8.9.1.

        * ace/config-sunos5.5.h: don't define ACE_HAS_XPG4_MULTIBYTE_CHAR
          with ghs, because its version 1.8.9 doesn't seem to support it.
          Also, with ghs, replaced the __ctype [] declaration with an
          #include of <stdlib.h>, because that's cleaner.  Finally,
          added ACE_HAS_BROKEN_ENUMS for ghs (1.8.9).

Thu Jan 07 09:05:27 1999  David L. Levine  <levine@cs.wustl.edu>

        * ace/config-sunos5.5.h: added ACE_HAS_SIG_C_FUNC with
          Sun CC >= 4.2.  Sun CC 5.0 needs it; Sun CC 4.2 doesn't
          object to it on SunOS 5.5.1, 5.6, and 5.7.  Thanks to
          Neil Cohen <nbc@aikisoft.com> for figuring out how to
          compile ACE with Sun CC 5.0.

        * include/makeinclude/platform_sunos5_sunc++.GNU: only
          use -features=castop and -rtti with Sun CC 4.2.  CC 5.0
          doesn't support them; I assume that it enables RTTI by
          default.  Thanks to Neil Cohen <nbc@aikisoft.com> for
          reporting this.

        * ACE-INSTALL.html: added ACE_HAS_REGEX #undef suggestion for
          shared lib link problems with egcs 1.1.x on Solaris 2.5.x.
          Thanks to Bob McWhirter <bob@werken.com> for reporting this
          problem.

Thu Jan 07 04:45:36 1999  Douglas C. Schmidt  <schmidt@cs.wustl.edu>

        * ACE version 4.6.12 released.

Thu Jan 07 04:01:08 1999  Irfan Pyarali  <irfan@cs.wustl.edu>

        * ace/Containers_T.cpp (max_size): No need to reallocate if the
          new size is the same.  Thanks to Mike Goldman <whig@by.net> for
          pointing this out.

        * ace/Strategies_T.cpp (check_hint_i and cleanup_hint_i):
          purge_i() should be called after the entry is closed.

        * Select_Reactor_Base.cpp (unbind): The next <max_handlep1_>
          should be calculated not only based on the <wait_set> but also
          the <suspend_set>.

          Thanks to Mark L. Boriack <mboriack@std.saic.com> for providing
          a description of this bug.

Wed Jan 06 21:42:11 1999  David L. Levine  <levine@cs.wustl.edu>

        * ace/OS.cpp (thr_create): moved #endif /* ACE_LACKS_SETDETACH */
          outside of }, to allow compilation on DU 4.0/cxx with DCE threads.

        * ace/config-osf1-4.0.h: with DCE threads only, added
          ACE_LACKS_THREAD_PROCESS_SCOPING.

Wed Jan 06 16:36:21 1999  David L. Levine  <schmidt@cs.wustl.edu>

        * ACE version 4.6.11 released.

Wed Jan 06 15:15:37 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

        * ace/Service_Config.cpp (process_directive,close): Moved the
          deletion of <svc_conf_file_queue_> into close method.
          Otherwise, there's no way to reconfigure the service
          configuration once it get started.  Thanks to Mike Goldman
          <whig@by.net> for reporting the problem.

Wed Jan  6 14:27:24 EST 1999  James CE Johnson  <jcej@lads.com

        * docs/tutorials/021/*:
        Completed the ACE_Malloc<> tutorial.  It could be a lot better
        than it is but I'm still new to ACE_Malloc...

        * docs/tutorials/021/online-tutorials.html:
        Added the link for #21

Tue Jan 05 22:55:07 1999  Irfan Pyarali  <irfan@cs.wustl.edu>

        * ace/Strategies_T.cpp (cleanup_hint_i and check_hint_i): Since we
          are holding the connection cache lock in these methods, the
          Svc_Handler should not call purge().  If it does, a deadlock
          will occur (unless we have recursive locks) since we are already
          holding the connection cache lock.  Therefore, we zero out the
          recycler before calling svc_handler->close() and purge out the
          svc_handler entry ourself (through purge_i()).

          Thanks to Mark L. Boriack <mboriack@std.saic.com> for providing
          a precise example illustrating this bug.

Tue Jan  5 18:48:47  James CE Johnson  <jcej@lads.com>

        * docs/tutorials/templates.html:
        Fixed a number of problems here.  Thanks to Amos Shapira
        <Amos_Shapira@icomverse.com> for pointing them out.

        * docs/tutorials/colorize:
        I pulled out the colorization code from combine into this
        standalone script.  One of these days I'll fix combine so that the
        code isn't duplicated...

Tue Jan  5 16:14:00 1999  Carlos O'Ryan  <coryan@cs.wustl.edu>

        * ace/config-irix6.x-common.h:
          Added ACE_TIMER_SKEW definition.

        * ace/FILE.cpp:
          Fixed the seek() prototype, is was declared as taking a "off_t"
          argument, but is was defined with a "long" argument,
          unfortunately in most platforms this was not a problem.

Tue Jan 05 13:15:35 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

        * ace/Service_Config.i (open): Be sure to pass down the
          <ignore_static_svcs> argument down to the real open method.

Tue Jan 05 08:08:06 1999  David L. Levine  <levine@cs.wustl.edu>

        * tests/MM_Shared_Memory_Test.cpp (spawn): fixed syntax error
          by moving ) to end of ACE_ERROR invocation.

        * ace/config-osf1-4.0.h: added 10 ms ACE_TIMER_SKEW, so that
          MT_Reactor_Timer_Test passes.

        * ace/config-lynxos.h: added 10 ms ACE_TIMER_SKEW, so that
          MT_Reactor_Timer_Test comes closer to passing.  It now
          chokes on the wait () near the end.

        * ace/OS.i (thr_getspecific): moved return 0 statement inside
          the Draft 7/STD #else block, to avoid warning about unreachable
          statement with Draft 4/6.

        * ace/config-osf1-4.0.h: with Draft 4 pthreads, added
          ACE_LACKS_READDIR_R.

        * tests/MT_Reactor_Timer_Test.cpp (main): added printout if
          wait () fails.

Tue Jan  5 00:59:04 1999  Douglas C. Schmidt  <schmidt@ace.cs.wustl.edu>

        * ace/Synch: Changed the remove() method of ACE_Thread_Mutex,
          ACE_Mutex, ACE_Semaphore, ACE_RW_Mutex, ACE_Event,
          ACE_Thread_Condition_Mutex, ACE_File_Lock, and
          ACE_Recursive_Thread_Mutex so that they check a flag to see if
          remove() has already been called.  If it has, then the remove()
          call does nothing.  This avoids nasty problems on platforms like
          Win95 that fail mysteriously if locks are destroyed multiple
          times.  Note that our solution isn't perfect since it won't
          protect against race conditions if multiple threads call
          remove() simultaneously.  However, if this happens it's a sign
          that the application was designed incorrectly...  Thanks to Paul
          Felix <pef@fluent.com> and Irfan for tracking this down.

        * ace/FILE: Added new methods, seek() and tell(), that are more
          intuitively named than the existing position() methods.  The
          position() methods have been retained for backwards
          compatiblity, but are now marked as deprecated.

        * ace/OS.h: Added a macro for ACE_DEFAULT_HTTP_PORT, which
          is 80 of course...

        * ace/INET_Addr.cpp: Make sure that addr_to_string() correctly
          checks the length of its buffer before doing a sprintf().  In
          addition, added an option to print the results in either ip-addr
          format or ip-name format.

        * ace/Mem_Map.cpp (map): where ace/Mem_Map.cpp mmap() will succeed
          if the length of the file mapping is 0, which will be the case
          if we've just created the file.  This was the wrong place to
          make this check.

        * tests/MM_Shared_Memory_Test.cpp: Cleaned up a few things
          in this test so that it'll be easier to maintain.

Mon Jan 04 08:45:45 1999  David L. Levine  <levine@cs.wustl.edu>

        * ace/config-sunos5.7.h: only use the g++ hacks if the
          g++ version is prior to 2.8.  Assume that later versions
          were built on SunOS 5.7 host, and/or don't have header
          files that are incompatible with the system headers.

        * bin/create_ace_build: changed symlink creation failure from
          error (with termination) to warning create_ace_build.

Mon Jan  4 03:15:57 1999  Douglas C. Schmidt  <schmidt@ace.cs.wustl.edu>

        * ace/Mem_Map: If we're remapping an already-mapped file
          (i.e., the base_addr_ != 0) && the addr parameter IS 0 (i.e.,
          the user isn't trying to respecify where to map this file), then
          we'll try to map over top of the existing region, which implies
          "MAP_FIXED".

Sun Jan 03 23:21:46 1999  Nanbor Wang  <nanbor@cs.wustl.edu>

        * ace/Service_Config.cpp (initialize): Both initialize methods now
          remove the service object from the repository if its init()
          method failed.

Sun Jan  3 14:39:49 1999  Douglas C. Schmidt  <schmidt@ace.cs.wustl.edu>

        * ace/ACE: Added a simple dirname() wrapper method.

        * ace/SString: The set(const char *, size_t, int = 1) and
          set(const char *, size_t) methods were inherently ambiguous.
          I've fixed this by changing removing the default parameters.
          Thanks to Irfan for suggesting this fix.

        * ace/Mem_Map.cpp: mmap() will fail if the length of the file
          mapping is 0, which will be the case if we've just created the
          file.  Therefore, don't even bother trying to mmap() in this
          case and return a successful result.

        * ace/FILE: Moved the get_remote_addr() and get_local_addr()
          methods from the FILE_IO class to the FILE class since that's
          more properly where they belong.  Also added a new remove()
          method that unlinks the file.

        * tests/Priority_Reactor_Test.cpp,
          tests/Conn_Test.cpp,
          tests/Notify_Performance_Test.cpp,
          Priority_Reactor_Test.cpp,
          Reactor_Performance_Test.cpp: Since the default is to skip argv0
          there's no sense explicitly saying this in the constructor of
          ACE_Get_Opt.

        * ace/FILE_Connector.h: Updated the comments to reflect recent
          changes.

        * ace/FILE.h: Added a new get_info() method that takes a reference
          to an ACE_FILE_Info rather than a pointer.  This is a cleaner
          API, IMHO.

        * ace/FILE_Connector.cpp (connect): When the <remote_sap> argument
          is ACE_Addr::sap_any, then connect will select a temporary
          filename using the new feature of ACE_FILE_Addr described below.

        * ace/FILE_Addr.cpp (set): When the address is ACE_Addr::sap_any,
          then we'll select a new temporary filename using the new
          ACE_DEFAULT_TEMP_FILE macro and ACE_OS::mktemp().

        * ace/OS.h: Added a new ACE_DEFAULT_TEMP_FILE macro that defaults
          to "/tmp/ace-fileXXXX" and is used by the ACE_FILE_Connector
          to create a temporary file when it's given an ACE_Addr::sap_any.

        * ace/FILE_Addr.h: The <set> accessor should return an int to be
          consistent.

        * ace/OS.h: Added a new ACE_sap_any_cast macro, which makes it
          much easier to cast the ACE_Addr::sap_any to the right subclass.

        * tests/Reactor_Performance_Test.cpp,
          tests/Reactor_Exceptions_Test.cpp,
          tests/Process_Strategy_Test.cpp,
          tests/Priority_Reactor_Test.cpp,
          tests/IOStream_Test.cpp,
          tests/Conn_Test.cpp: Replaced all the C-style casts for
          ACE_Addr::sap_any with the new ACE_sap_any_cast macro.

Sun Jan 03 21:48:37 1999  Irfan Pyarali  <irfan@cs.wustl.edu>

        * ace/Asynch_Acceptor.cpp: Since this is a template file, I
          removed it from the ace library build on NT.

        * ace/Asynch_Acceptor.cpp (open): Changed C-style cast to use the
          new ACE_sap_any_cast macro from Dr. Schmidt.

Sun Jan 03 09:12:23 1999  David L. Levine  <levine@cs.wustl.edu>

        * ace/config-sunos5.5.h: with GreenHills, added
          ACE_LACKS_LINEBUFFERED_STREAMBUF and ACE_LACKS_SIGNED_CHAR.

        * ace/Basic_Types.* (ACE_U_LongLong): unioned the data members
          with a double, to ensure alignment on 8-byte boundaries.  It
          could matter when an ACE_U_LongLong is used as a true 8-byte
          quantity, such as in a call to Solaris ::gethrtime ().

        * tests/Time_Value_Test.cpp: replaced remaining #ifdef
          ACE_HAS_STHREADS with #ifdef sun.

Sat Jan  2 23:48:44 1999  James CE Johnson  <jcej@chiroptera.tragus.org>

        * docs/ACE-tutorials.html
        * docs/tutorials/guide-tutorials.html
        * docs/tutorials/new-tutorials.html
        * docs/tutorials/online-tutorials.html
        Converted some of the absolute URLs to relative.  I only have the
        docs/tutorials tree here at home so I couldn't do any more than
        these.

        * docs/tutorials/021/*
        New tutorial on ACE_Malloc<>.  I haven't begun the documentation
        yet but wanted to get the code commited so it doesn't get lost.

Sat Jan 02 08:34:27 1999  David L. Levine  <levine@cs.wustl.edu>

        * examples/ASX/Event_Server/Event_Server/event_server.cpp
          (handle_input),
          examples/Reactor/Misc/test_early_timeouts.cpp (main):
          added ACE_UNUSED_ARG (n), to avoid compile warning with ACE_NDEBUG.

        * performance-tests/Misc/test_naming.cpp (find): moved function
          call w/side effect out of ACE_ASSERT.  Thanks to DU 4.0 cxx
          for noticiing it, with ACE_NDEBUG.

Fri Jan 01 18:57:22 1999  David L. Levine  <levine@cs.wustl.edu>

        * ace/Containers_T.i (ACE_Array,ACE_Array::operator=):
          added missing ACE_INLINE's.

        * ace/OS.{h,cpp} (ACE_TSS_Info::operator{==,!=}): added
          constness.  It's necessary with the ACE_Array changes,
          with ACE_HAS_TSS_EMULATION.

Fri Jan  1 12:22:14 1999  Carlos O'Ryan  <coryan@cs.wustl.edu>

        * ace/Containers_T.h:
        * ace/Containers_T.i:
        * ace/Containers_T.cpp:
        * bin/Array_Helper:
        * bin/Array.pl:
          Moved most of the implementation of ACE_Array into
          ACE_Array_Base this class does not require == or != operators
          defined for the template parameter.
          I also added methods to access and manipulate the capacity of
          the array (max_size) as opposed to always changing its size (the
          portion actually used).

        * ace/Object_Manager.cpp:
        * ace/Template_Instantiations.cpp:
        * ace/OS.cpp:
        * apps/JAWS/clients/Caching/URL_Properties.cpp:
          Fixed ACE_Array instantiations.