Editions of the IKR SimLib

The IKR SimLib is currently in a transition phase and therefore several release branches exist:

• The IKR SimLib C++ Edition Version 2.6 is very stable and used in many existing projects. In the future, this release will receive only bug fixes.
• The IKR SimLib C++ Edition Version 2.7 is rather new and contains a new feature compared to the C++ Edition Version 2.6: This release contains logic to run batches on different processor cores in parallel (called "batch parallelization"). In computing environments with a lot of processor cores this allows to reduce the duration of (selected) simulations. In addition, this release supports the separate simulation and combined evaluation of multiple instances of the same simulation model with different random number seeds (a simulation method called "independent replication"). The output file format was changed to XML to better reflect the hierarchical structure of the simulation model. The IKR SimLib C++ Edition Version 2.7 will receive only bug fixes. We do not add new features because we consider C++ as a dead end.
• The new IKR SimLib Java Edition is currently in Version 2.8. This branch will receive bug fixes as well as new features. We use it in most of our current and all of our new projects. The current release 2.8.0 has all the features of the C++ Edition Version 2.7 (including the batch parallelization and the new output file format) and some additional features like packing of output files with bzip2-/lzma-compressors.

The new XML based output file format is structurally equivalent to the formerly used flat file format. Due to XML the extraction of values is not as simple as before but should be no problem with appropriate tools.

Which IKR SimLib to choose?

If you are unsure which edition and/or revision to select, consider the following:

• If you have a (not small) existing project based on the IKR SimLib C++ Edition in Version 2.6 and you see no benefits from using your processing resources in parallel, then stay with the C++ Edition Version 2.6 - there is no need to leave this stable release.
• If you start with a new project and are not urged to use C++ then select the Java Edition Version 2.8. This release is future oriented and proved to be stable.
• If you have existing projects based on the C++ Edition and expect significant benefits from batch parallelization then consider updating to the C++ Edition Version 2.7.
• If you have small existing projects based on the C++ Edition then consider updating to the Java Edition. We have ported several small- to medium-sized projects and found porting these projects from C++ to Java rather simple and more than worth the effort. We experienced a large improvement in development speed and observed no drawbacks at all.