Laboratory for Software Design Laboratory for Software Design

A Paper on Candoia Platform and Ecosystem will Appear at MSR 2017, a conference co-located with ICSE 2017

March 14, 2017

Links: Candoia | MSR 2017 | ICSE 2017

A paper on the Candoia Platform and Ecosystem has been accepted at the 14th International Conference on Mining Software Repositories (MSR 2017), a conference co-located with ICSE 2017 at Buenos Aires, Argentina, 20-21 May, 2017. Main goal of the platform is to bridge the gap between research and practice thereby easing the process of transitioning software engineering research results to industry.

Over the last decade, mining and understanding software repositories (MSR) research has shown significant advances in several critical software engineering (SE) areas. Key successful research sub-areas in MSR are: defect prediction, bug fixing effort estimation and suggesting fixing experts, social network analysis for software programming pattern discovery, specification mining, etc. However, this research has not yet focussed on the question of widely-distributing MSR tools. There are frameworks and platforms that aim to ease deployment of program analysis tools; however, their focus is not on MSR.

There are at least three major technical challenges in transitioning MSR research to practice. First, most researchers, and adequately so, focus on realizing their research as a program suitable for their own experiments — there is a significant cost to converting these prototypes into software intended for wide usage. Developing MSR tools as plugins for platforms like Eclipse can help with that, however, the level of abstraction that these platforms provide is still too low for MSR tools. Second, in order to be broadly applicable a MSR research prototype must integrate with a variety of tools — version control systems (VCS) such as CVS, SVN, Git, etc., bug databases such as Bugzilla, Issues, Jira, etc., forges such as SF.net, GitHub, Bitbucket, etc., programming language such as Java, Javascript, PhP, etc. Expecting such broad applicability from every tool meant to evaluate research hypotheses may not be reasonable, and may substantially increase the cost of scientific research in this area. Third, usage scenarios that the researchers may have used in their experiments may not exactly match the need for all users of their tool, requiring users to slightly customize the tools to fit their purpose. If the implementation of the tool is too complex that may challenge the resolve of users to customize the tool for their own projects. These three challenges substantially increase the cost of research-to-practice transition in this area.

To solve these problems, we have designed a platform for realizing MSR tools that we call Candoia (pronounced can-doy-uh). Candoia is to MSR tools as Android and iOS are to mobile applications. Like these platforms, Candoia provides suitable abstractions for building MSR tools and handles the details of integration with VCS, bug databases, language parsers, visualization, etc. Researchers prototype their research as Candoia apps and can publish them. Practitioners install Candoia platform and can browse through available apps, download them, and if necessary customize them for their own needs.
We have implemented Candoia and make it available for download at (http://candoia.org). We have also created an exchange for Candoia apps, and populated it with an initial set of apps. The Candoia platform can interact with the exchange to fetch published apps.

Our MSR 2017 paper will desribe all of these innovations.

More information about the project.

  1. Candoia Web-site https://candoia.org.
  2. Candoia GitHub Project https://github.com/candoia.
  3. Contact Dr. Hridesh Rajan for more information via e-mail hridesh@iastate.edu.