Gin

Abstract

Genetic improvement (GI) is a young ield of research on the cusp of transforming software development. GI uses search to improve existing software. Researchers have already shown that GI can improve human-written code, ranging from program repair to optimising run-time, from reducing energy-consumption to the transplantation of new functionality. Much remains to be done. The cost of re-implementing GI to investigate new approaches is hindering progress. Therefore, we present Gin, an extensible and modiiable toolbox for GI experimentation, with a novel combination of features. Instantiated in Java and targeting the Java ecosystem, Gin automatically transforms, builds, and tests Java projects. Out of the box, Gin supports automated test-generation and source code proiling. We show, through examples and a case study, how Gin facilitates experimentation and will speed innovation in GI. CCS CONCEPTS• Software and its engineering → Software notations and tools; Search-based software engineering;

Comparing Apples and Oranges? The Influence of Java Profiler Choice on the Identification of Hot Methods for Genetic Improvement

“…There are many profiling tools available for Java. Gin versions up to v2.0 (White, 2017;Brownlee et al, 2019) made use of the HPROF tool (Oracle Systems, 2020) included with JDK versions up to 8. Watkinson and Brownlee (2023) summarised the desirable qualitative properties of several current profiling tools with a view to upgrading Gin for newer versions of Java.…”

“…Instead, profiling appears to be often done by the respective team on an ad-hoc basis and with a variety of tools due to the targeted applications and objectives. For example, Haraldsson et al (2017) (White, 2017;Brownlee et al, 2019), which forms the focus of the present study, was developed with the aim to facilitate GI research. Gin is opensource, meaning researchers can build upon the existing code, and as far as possible makes use of multiple existing tools such as EvoSuite (Fraser and Arcuri, 2011) for automated test creation.…”

“…In choosing the regions of code that GI will target for improvement, the profiler plays an important role, and variations in a profiler's output will greatly impact the performance of the GI search, motivating us to seek better understanding of the consistency and behaviour of profilers. The Gin open source toolbox for GI research in Java, originally proposed by White (2017) and further developed by Brownlee et al (2019), was recently upgraded by Watkinson and Brownlee (2023) to support Java 17 and part of that process switched from the retired HPROF profiler to the more recent Java Flight Recorder (JFR). Watkinson and Brownlee (2023) outlined the initial process of changing the profiler in Gin and explored the difference in profiling outputs between HPROF under Java 8 and JFR under Java 9.…”

Impact of Test Suite Coverage on Overfitting in Genetic Improvement of Software

“…We used the genetic improvement toolbox Gin v2.0 [6] in our experiments. Gin fulfills all our requirements: it is open-source, targets Java programs, uses runtime improvement as fitness by default, and its second release provides integration with EvoSuite [11].…”

“…Unlike in the automated program repair field, there is no standard benchmark for runtime improvement using GI. We also require the programs to be relatively small, so we could run thousands of experiments in reasonable time and avoid the, often very costly, profiling stage of the GI process [6], targeting the whole software instead. With those restrictions in mind, we chose to use the set of 9 sort algorithms and the triangle example provided with the first release of Gin [23] in our study, for which improvements have previously been found using GI [9].…”