adapt to SWIG version 3 (define Python package)
remove unused includes (iostream)
adapted SWIG tests on Windows for Python >= 3.8 (new approach for DLL loading)
added CMake target for deploying Java wrapper to remote repository
use more precise maven classifier (with/without, sundials, OS, 32/64-bit)
fixed Sundials support for Swig wrappers
added custom command to install shared libraries to local Maven repository; moved all custom commands related to Maven to main CMakeLists.txt
changed Java package to 'at.ac.ait.fmipp.imp' and 'at.ac.ait.fmipp.exp'; changed Java JAR file names to 'fmippim.jar' and 'fmippex.jar'; updated tests accordingly; changed to new SWIG version (swig_add_library,CMP0078,CMP0086); added Maven install target;
Deprecation causes build failure with latest version of SUNDIALS
entry error in example
Missing Example for FMUCo-simulationV01 / V02
fix warning for previous commit (comparison of 'int' and 'unsigned int'); fixed problems of previous commit for compiling with MinGW (added missing compiler flags);
added functions retrieving value refs and names of state vector and derivatives; added function to enter event mode for FMU ME v2.0;
define default values in IncrementalFMU's sync function inputs
Thank you for your speedy reply, I didn't realize that I had duplicated another issue. Timo van Opstal Design Engineer Sioux LIME limebv.nl 040-7516129 Esp 405, 5633AJ Eindhoven On 17 Jan 2019, at 10:12, Edmund Widl widle@users.sourceforge.net wrote: Dear Timo, Thank you for pointing this out. Your observation is basically a reflection of the open tickets for the FMI++ library. These tickets are mostly related to edge cases, which in most applications do not pose a problem. However, to be able to...
Dear Timo, Thank you for pointing this out. Your observation is basically a reflection of the open tickets for the FMI++ library. These tickets are mostly related to edge cases, which in most applications do not pose a problem. However, to be able to solve them we included dedicated tests. I hope that I will have the time to fix the related tickets soon ... Nevertheless, as said above, in most applications these bugs do not really pose a problem. Therefore, for the time being I suggest to simply...
Broken test
bug fix: proper conversion of flag 'loggingOn' when using function 'setDebugFlag' of front-end
bug fix of previous commit: use 'varname' instead of '#varname' in macros
add macros for defining backend variables with specific name (instead of using C++ variable name)
added flag to turn on/off static linking for boost libraries
removed MATLAB scripts (moved to separate repository)
do not compile function 'giveZero()' into extra shared library on Linux'
Merge branch 'master' of https://git.code.sf.net/p/fmipp/code
removed redundancies in CMake config files
increase tolerance for test slightly to work with MSVC 64-bit
look for external packages (Java, Python, etc.) used by SWIG in main CMake script; do not look for Java when running tests but not building Java bindings;
in case the persmissions of the backend application can not be changed just give a warning, but do not exit
Use 'cygwin32' / 'cygwin64' as binary directory name when compiling under Cygwin
add function loggingOn() to BackEndApplicationBase; lots of cosmetic changes;
fixed indentation
Merge branch 'master' of ssh://git.code.sf.net/p/fmipp/code
added debug message in case parsing of additional arguments fails; setter functions: check if variable is scalar parameter according to FMI 1.0;
Only run python and java tests if we build the swig wrappers for them
Moving to github would be a great step to welcome external contributions. Github or gitlab or bitbucklet, they all make it very easy to fork a project, craft a set of patches and submit them to the original author. SourceForge makes the life more complicated for external contributors that just pass by, and don't want to become a regular contributor. That being said, it's not because you go to github/gitlab/bitbucklet that armies of contributors studendly appear, I know that. But at least, it reduces...
Why? Where's the rationale for such a decision? Currently Edmund is doing almost 100% of the work, so why move to a platform that's tailored to large development communities? Just wondering... IMHO, work (especially such that is imposed on others) should always be well motivated :-)
removed Python scripts
removed superfluous code checking if interface is active
moved configuration for BOOST linking (static/dynamic) to main CMake file
added two new compilation targets (used by other projects to generate FMUs for Co-Simulation)
Yes please! github, gitlab, bitbucklet, anything but not sourcefoge! Thanks
changed timeout for tests from 10 seconds to 10 minutes
prevent test on Linux to fail because of SIGCHLD signals (which come from FMU implementation, have nothing to do with the tests at hand)
fixed test name in set_properties(...) for Linux Java tests
add timeout of 60 seconds to tests
added internal flag for FMI version to backend/frontend; set causality of parameters in dependence on FMI version
start external applications in their own process group (LINUX); terminate all child processes of an application (LINUX: via process group, WINWOWS: via parent ID)
in case no entry point is provided, use path of executable as working directory
improved backend's functionality to write input/output variable names and parameter names to file (write to JSON file, do not stop execution with exit status indicating failure)
minor changes to fix compiler warnings (Linux GCC)
enable secific shared memory segment names; enable generation of shared memory segment names from parent PID;
check status of FMU loading before checking status of FMU type
fmi_2_0::FMUModelExchange: Enter the event mode on setting discrete inputs
fixed sending of debug messages
added include directories needed by MINGW
fixed bug: contructor was given wrong 4th argument
removed instances of std:: (namespace is used anyway)
FMUCoSimulation_v2: use internal time for doStep; InterpolatingFixedStepSizeFMU: add iterateOnce(); FixedStepSizeFMU: cosmetic changes;
improved handling of time resolution for communication points
updated handling of model description for FMI CS: new structure due to FMI 2.0 changes, retrieval of capability flags
add tests
added support for FMI CS 2.0 for utility classes
removed tool-specific parts of export package
FIX: Broken isnan(...)
ADD: == and != for Integrator::Properties
Support for Integrator::Properties in FMUModelExchange and IncrementalFMU
Merge branch 'issue-16-integrator-properties'
ADD: Integrator::Property support in IncrementalFMU (#16)
Support via void FMUModelExchangeBase::setIntegratorProperties( Integrator::Properties& properties ) partially available. IncrementalFMU support is still missing
Support for Integrator::Properties in FMUModelExchange and IncrementalFMU
fix model description to allow running test under Linux (removed '.exe' from executable name)
improved FMI CS 2.0 support of import package; added new tests accordingly;
fixed typo in model description file
updated model description for testing
added MINGW workaround concerning boost filesystem library (undefined reference to boost::filesystem::detail::copy_file)
Merge branch 'master' of ssh://git.code.sf.net/p/fmipp/code
export package bow uses almost same structure for storing data in the model description's vendor annotations
added FMI 2.0 support for export package
MOD: Changed the update policy of der(x)
2.0 ME only FMU fails to load if co-simulation functions are missing
Merge branch 'issue-9-me-2.0-only'
FIX: Selectively load FMI 2.0 functions (#9)
FIX: Spurious 'static' statements
MOD: Removed Duplicated Code
MOD: Removes duplicate DLL/SO access code
2.0 ME only FMU fails to load if co-simulation functions are missing
Merge branch 'master' of ssh://git.code.sf.net/p/fmipp/code
class FMIComponentBackEnd initializes now the variables corresponding to 'stop_time' and 'stop_time_defined'
MOD: Disable UBLAS runtime type checks
Automatically deduce FMU model identifier before loading
FIX: Missing include directive
Merge branch 'issue-15-automatic-type-name'
ADD: Getter function for IncrementalFMU::timeDiffResolution_
ADD: C'tor for IncrementalFMU which takes the model identifier and no URL
MOD: Made ModelManager::getTypeOfLoadedModel(...) public
ADD: Automatic name deduction and unload function
FIX: Flaw in ModelDescription::hasModelIdentifier(...)
MOD: Refactored ModelManager::loadFMU(...)
ADD: Function stub and test cases for #15
Automatically deduce FMU model identifier before loading