SCC reuses the first parse and the second parse of SVM to construct the same internal structures. It synthesize code in multiple target languages from those internal structures. Currently it supports the following target languages: Java, C++, C# and Python.
The models are optimized before code synthesis. For example, an unique integer number is assigned to each event, which is used as the internal representation of the event instead of its string name. The names of states are also translated into integer numbers. This speeds up the execution by transforming string comparison to integer comparison.
Compared to the simulation in SVM, execution of the code generated by SCC is much more efficient. However, the code must be compiled by appropriate compiler for the target language before running. Because the code is statically generated, the flexibility of dynamically modifying its behavior by means of macro redefinition is lost.