[ace-users] ACE+TAO+CIAO x.5.9 now available

ciju john john_c at ociweb.com
Tue Jun 26 17:11:15 CDT 2007

Hi all,

Thanks to the substantial efforts by many developers, testers, and
users, we are pleased to announce the latest ACE+TAO+CIAO beta.  It is
available from the same download location as per the previous release at


under the heading "Latest Beta Kit".

The doxygen documentation for the beta is also available. Included with
this beta, besides the packages combined of sources and generate
makefiles, we also provide source only packages for the group of users
which use MPC to generate their own makefiles.

We encourage you to download the new beta, use it with your applications,
and let us know if you encounter any problems. Please use the:


so that we have the proper version/platform/compiler/options you're using
to report problems. We also request that you take a look at:


for the status of various ongoing projects at the DOC groups of WashU,
UCI, and Vanderbilt to move ACE+TAO+CIAO forward.  Overviews of our
recent progress and upcoming plans are available at:


The overall success rates for the test results gathered from all our
daily builds is 97% for the ACE tests and 94% for the TAO tests.  Please


for a summary of these tests.  As always, we appreciate your help in
resolving any remaining problems.  Please see:


for an overview of our bug fixing policies.

The following is a list of key changes since the last release.  Please
refer to the corresponding ChangeLog files, NEWS files or Bugzilla
entries for more details on the problem and the nature of fixes applied.

Changes to ACE

. Use Intel C++ specific optimizations for Linux on IA64

. Improved support for ACE_OS::fgetc. Added support for ACE_OS::fputc,
  ACE_OS::getc, ACE_OS::putc and ACE_OS::ungetc.

. Added support for ACE_OS::log2(double) and improved support for

. Shared library builds on AIX now produce a libxxx.so file instead of the
  previous practice of producing libxxx.a(shr.o).

. GCC 4.1.2 that comes with Fedora 7 seems to have a fix for the visibility
  attribute we use for the singletons. F7 users will therefore need to
  define the following in your config.h file.

. Fixed (rare) problem in TP_Reactor where incorrect event handler was

. Reduced footprint on some platforms, particularly those that use
  g++ >= 3.3.

Changes to TAO

. When using AMI collocated in case of exceptions they are deliverd
  to the reply handler instead of passed back to the caller

. Added support for IPv6 multicast addresses when federating RTEvent

. Fixed a bug in the IDL compiler's handling of octet constants where
  the rhs consists of integer literals and infix operators (Bugzilla 2944).

. TAO_IDL enhancements for built in sequence support in TAO DDS.

. Provide support for TAO DDS zero-copy read native types.

. TAO_IDL fix for lock-up and incorrect error message generation for
  missing definitions within local modules. This fix also stops the
  erronious look-up in a parent scoped module (with the same name as
  a locally scoped module) that should have been hidden by the local
  scope definition.

. The TAO_IORManip library now has a filter class that allows users to
  create new object references based on existing multi-profile object
  references by filtering out profiles using user defined criteria.  The
  use of -ORBUseSharedProfile 0 is required for this to function.

. A problem in TAO_Profile::create_tagged_profile was fixed. This problem
  triggered only when MIOP in multi-threaded application was used.

. Added IPv6 support in MIOP so that IPv6 multicast addresses can be used in
  addition to those IPv4 class D addresses. DSCP support is implemented in
  MIOP as well. However, since MIOP only allows one way communication then it
  makes sense to use DSCP on the client side only.

Changes to CIAO

. Added a new deployment algorithm to DAnCE for optimization of large
 scale deployment. In this algorithm, the number of threads spawned is
 based on the deployment plan, i.e, by parsing the deployment plan
 information, DAnCE decides how many threads to spawn (one thread per
 node). This algorithm knows how to "initialize" each thread since
 each thread will have different execution context, i.e., which
 components to deploy to which node, component configuration, etc.

. Added a NA component server callback wait strategy in NAM, which uses
 conditional variable to signal the NA call back so it can work with
 multi-threaded configuration, such as thread-per-connection mode. The
 original implementation uses the main thread to run the ORB event
 loop, which will not work for multi-threaded environment.

Bugzilla Bug Fixes

ACE bugs fixed:
475, 2710, 1563, 2453, 2895, 2916, 2919, 2922, 2923, 2948

TAO bugs fixed:
2377, 2710, 2473, 2888, 2909, 2921, 2929

More information about the Ace-users mailing list