Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
branch: master
Commits on Feb 19, 2015
  1. Tim Pokorny

    Merge pull request #127 from michaelrfraser/bug/port124-winHLA13Tests…

    timpokorny authored
    …DontCompile
    
    PORT-124 hla13 tests dont compile under Windows
  2. Tim Pokorny

    Merge pull request #128 from michaelrfraser/bug/port126-crashedFedera…

    timpokorny authored
    …teDoesNotResignFromFederation
    
    PORT-126 Resign message now forwarded to LRC when a federate crash is confirmed
  3. Michael Fraser

    PORT-126 Resign message now forwarded to LRC when fed crash is confirmed

    michaelrfraser authored
    - Confirmed that FederateListener.acceptView(View) is called when the
      underlying JGroups mechanism confirms (not suspects) that a node has
      crashed
    - Previously FederateListener.acceptView(View) was synthesising a
      ResignFederation message for each crashed federate, but never
      dispatching it to the LRC for processing.
    - The resign messages are now forwarded on with a call to this.receive()
Commits on Feb 18, 2015
  1. Michael Fraser

    CNR-124 Unit tests now compile and pass under Windows

    michaelrfraser authored
    - DEBUG linker flag is now set when the test suite is compiled
    - Test suite now links against a debug version of cppunit
    - Heap corruption was happening in the test utility class ActiveSR due
      when the char* label memory was deleted at some stage. Converted it into
      a std::string so memory management and got rid of all the new/delete
      calls
    - Fixed an iterator bug in Test13FederateAmbassador::fetchInteraction()
      * vector->erase() modifies the iterator state, so returning the
        dereferenced value of it in the next statement actually returns the
        next item in the vector
      * if the next item is end(), then the Microsoft version of STL throws a
        CRT exception saying the iterator can't be dereferenced
    - Fixed a bug in
      RequestUpdateWithRegionTest::testRequestUpdateWithRegion() where end()
      was being called on a deleted map
Commits on Feb 17, 2015
  1. Tim Pokorny

    Merge pull request #125 from michaelrfraser/task/port99-cppDeprecated…

    timpokorny authored
    …StringXforms
    
    PORT-99 Fixed missed deprecated string transform in Runtime.cpp
  2. Michael Fraser

    PORT-124 Added vc10 sub-dir to PATH when executing hla13 tests

    michaelrfraser authored
    - On my development machine, I didn't get any linker errors, however the
      test suite failed to run.
    - Found that the exec task that runs the hla13 test suite was setting the
      PATH env variable to ${hla13.complete.dir}, however the actual compiled
      lib exists at ${hla13.complete.dir}\vc10
    - Added vc10 subdir to the PATH variable and the suite now executes
    - Now getting a crash when running the test suite in the middle of
      DDMSupportServicesTest. Stack trace doesn't look to have any RTI or test
      suite code in it, so it may be happening when tearing down a test case
Commits on Feb 12, 2015
  1. Michael Fraser

    PORT-99 Fixed missed deprecated string transform in Runtime.cpp

    michaelrfraser authored
    - Modified getUnixPath function to exclusively use strings instead of
      char*
  2. Tim Pokorny

    Merge pull request #123 from michaelrfraser/task/port99-cppDeprecated…

    timpokorny authored
    …StringXforms
    
    PORT-99 HLA13 cpp tests no longer cause deprecated string transform warnings
Commits on Feb 7, 2015
  1. Tim Pokorny

    Fix #108 - Portico will no longer respond to requests to update MOM

    timpokorny authored
    attributes when the MOM is disabled. Put checks in the incoming message
    handlers for those calls to see if the MOM is enabled or not. If not,
    the call will return without doing anything.
Commits on Feb 5, 2015
  1. Michael Fraser

    PORT-99 HLA13 cpp tests no longer cause deprctd string transform warns

    michaelrfraser authored
    - Problem was due to c++11 deprecating the assignment of a non-const
      char* to a literal
    - char* params in all test related functions have been analysed and
      converted to const char* where appropriate
    - The MAP_LTSTR_CLEANUP in Common.h has been modified to work with a
      map<const char*,const char*>
    - The SET_LTSTR_CLEANUP in Common.h has been modified to work with a
      set<const char*>
    - Modified Test13Federate::quickReflectFail to handle situation where
      the provided const char* tag parameter is NULL
    - Modified Test13Federate::quickSendFail to handle situation where the
      provided const char* tag parameter is NULL
    - Test13Federate::announced is now a map<const char*,const char*>
    - Test13Federate::synchronized is now a set<const char*>
    - The char* saveLabel member variable in FederationRestoreTest has been
      changed to a static const char* variable
    - The char* saveLabel member variable in FederationSaveTest has been
      changed to a static const char* variable
    - HLA13 and Test13 code no longer compiles with the deprecated string
      warning disabled
  2. Tim Pokorny

    Fix #54 - I believe the cause of the NPE was that the local federate was

    timpokorny authored
    resigning, which was stripping the LRC off the message receiver, but
    that a separate thread was still processing a message, and when it got
    to hand off to the LRC, found that it was null. I've put both the
    receive and lifecycle methods into "synchronized" so they can't be run
    parallel to one another now which should clear it up. Haven't been able
    to reproduce the problem since, but will keep an eye out.
Commits on Feb 2, 2015
  1. Tim Pokorny

    Changed the 1516e RTIamb so that it no longer logs a message about every

    timpokorny authored
    exception at the ERROR level. Messages are now logged at debug. Just
    because there is an exception doesn't mean there is an error. There are
    many sitautions where we may expect an exception (such as trying to
    destroy a federation that is still in use when the local federate shuts
    down). Exposing this as an error message means it is printed in the
    console for every user, all the time. Dropped the level to DEBUG so that
    it is still logged, but you have to want to see more information to get
    it.
  2. Tim Pokorny

    Bumped the "MOM not supported in 1516e" warning message down to info. It

    timpokorny authored
    was just too nagging to have it pop up on every. single. execution. It's
    a message for the developer, and I am making every *user* look at it.
    Poor UX, Tim. Poor UX.
    
    Now falls below the default WARN level, but can be seen easily by setting
    the level to INFO or lower.
Commits on Jan 31, 2015
  1. Tim Pokorny

    Updated the logging configurator so that the log file does not roll

    timpokorny authored
    over. It will now be overwritten each time you do a simulation run.
Commits on Jan 30, 2015
  1. Tim Pokorny
Commits on Jan 28, 2015
  1. Tim Pokorny
  2. Tim Pokorny

    Fix #76 - Fixed typo in the service not supported exceptions generated

    timpokorny authored
    by calls to the save/restore methods in 1516e.
Commits on Jan 24, 2015
  1. Tim Pokorny

    Working #120: Added proper clone() support to UpdateAttributes. And by

    timpokorny authored
    "proper", I mean enough to support the JVM. It still isn't a true clone
    in the sense that all the values are not copied. But as most attributes
    of message classes are read-only on the incoming handlers, it is safe to
    only have a shallow clone. Awesome.
Commits on Jan 23, 2015
  1. Tim Pokorny

    Fix #63. Turns out I was overwriting the JGroups configuration by just

    timpokorny authored
    setting up my own Executor, with its own arbitrary settings, in code.
    This was falling back to the default rejection handler of Abort, which
    was causing the exceptions. Fixed now so that the executor will use the
    jgroups configured executor and the setting of "Run" will flow through.
Commits on Jan 20, 2015
  1. Tim Pokorny

    Fix #118 Calling disconnect() now properly shuts the immediate callback

    timpokorny authored
    dispatcher down if it is running. Also made the dispatcher thread a
    daemon so that it will never hold a process open by itself.
  2. Tim Pokorny

    Fix #67: Sync point registration failure message due to a point already

    timpokorny authored
    being registered is now only printed once and is done at DEBUG level.
Commits on Jan 15, 2015
  1. Tim Pokorny
Commits on Jan 14, 2015
  1. Tim Pokorny

    Fix #64 - Added a detail mode that can be set to full (default) or

    timpokorny authored
    summary. In summary mode, logs for individual messages will not be
    stored, regardless of filter settings.
Commits on Jan 12, 2015
  1. Tim Pokorny

    Updated the default RID and JGroups config to:

    timpokorny authored
       - Expose the flow-control ratio setting
       - Fix a typo in the RID for the FRAG2 settings
       - Rationalize the documentation down a bit to focus on things users
    	 may actually use
  2. Tim Pokorny
Commits on Jan 10, 2015
  1. Tim Pokorny

    Tweaked the standard RID and JGroups configuration files to maintain

    timpokorny authored
    some more sensible defaults.
  2. Tim Pokorny
Commits on Jan 8, 2015
  1. Tim Pokorny
  2. Tim Pokorny

    Merge pull request #61 from openlvc/port186-hla13-64bit-library-name

    timpokorny authored
    HLA 1.3 64-bit DLLs now named with suffix _64
  3. Tim Pokorny

    PORT-186 #59 Updated the build system to produce HLA 1.3 binaries with

    timpokorny authored
    the suffix "_64" rather than just "64". Also updated the native library
    loader (which loads the native libraries into the java space to support
    calling back across the boundary for C++ federates). Tested and working
    on the Mac. Travis will test Linux and I'll manually test Windows.
  4. Tim Pokorny
  5. Tim Pokorny

    Updated the JGroups Auditor to use a straight file logger (rather than a

    timpokorny authored
    rolling one) so that we always get one big file. Also fixed a small
    typo.
Commits on Jan 7, 2015
  1. Tim Pokorny

    Merge pull request #60 from openlvc/port187-audit-log-update

    timpokorny authored
    Audit Logging Extensions (FOM-Specific Info and Filtering)
  2. Tim Pokorny

    PORT-187 #58 Updated the auditor to support filtering on direction,

    timpokorny authored
    message type and fom class type
  3. Tim Pokorny

    PORT-187 #58 Cleaned up fom-specific message summary logging as well as

    timpokorny authored
    some of the core logging code. Changed the ordering of the summary
    tables to order by messages with higher counts (rather than higher total
    send/receive size).
Something went wrong with that request. Please try again.