Javolution

Javolution is a real-time library aiming to make Java or Java-Like/C++ applications faster and more time predictable. Indeed, time-predictability can easily be ruined by the use of the standard library (lazy initialization, array resizing, etc.) which is not acceptable for safety-critical systems. The open source Javolution library addresses these concerns^[1] for the Java platform and native applications. It provides numerous high-performance classes and utilities useful to non real-time applications as well. Such as:

• Collections^[2] classes, supporting custom views,^[3] closure-based iterations, map-reduce paradigm, parallel computing, etc.
• Worst-case execution time behavior documented using Realtime^[4] Java annotations
• Fractal structures^[5] to maintain high-performance regardless of the size of the data
• OSGi contexts^[6] allowing cross cutting concerns (concurrency, logging, security, ...) to be addressed at run-time through OSGi published services without polluting the application code (separation of concerns)
• Algorithmic parallel computing support with concurrent contexts^[7] and general-purpose computing on graphics processing units with compute context^[8]
• Struct/Union^[9] base classes for direct interfacing with native applications
• Perfometer^[10] utility class to measure worst-case execution time with high precision
• XML Marshalling/unmarshalling facility^[11] capable of direct serialization/deserialization of existing classes (no intermediate data structure required)
• StAX-like XML reader/writer^[12] which does not require object creation (such as String) and consequently faster than standard StAX
• Simple yet powerful configuration management^[13] for your application.

Since version 6.0, Javolution makes it easy to port^[14] any Java source code to C++ for cross-platform native compilation. OSGi and JUnit have been ported and are included with the core C++ distribution.

All modules (Java and native) are built using maven. Javolution can be used^[15] either as an OSGi bundle (preferred) or as a standalone library.