Laboratory of Software Comprehension, Analytics and Mining

Stic AmSud 2015 Project

Mining software repositories to instantiate software frameworks and react to API changes

 

Abstract:

Software systems are constantly getting larger and more complex. Nowadays it is not enough to reason about individual systems rather we have to examine entire ecosystems of software. The complexity of understanding and maintaining such ecosystems is mainly driven by the number of dependencies they have to other systems such as frameworks and libraries. As these external frameworks are beyond the control of the system developers, they can change in unforeseen ways, requiring developers to adapt their system to changes that occurred to an API of either a framework or a library. Also the initial usage of a new framework or library can impose a challenge as they are often not well documented.

In this project we aim at advising developers on how to instantiate a framework or library, in particular after significant changes and updates to its API. For this we mine large repositories of existing software systems to locate examples of API usages with the goal to characterize different usages. This will enable us to recommend to developers how to instantiate the framework for their particular use case at hand. We plan to integrate directly in the development environment recommendations based on existing examples of framework instantiations. This will allow developers to quickly find out how to use a framework or library to solve a particular problem while writing code.

This project will focus on software written in Java or Smalltalk, hence analyzing large repositories with systems written in these languages. As a cornerstone of this project we reuse existing tools and platforms we developed to efficiently analyze large-scale software repositories containing thousands of systems with millions of lines of code. 

 

Members:

Chile

David Röthlisberger (international coordinator), researcher, Universidad Diego Portales

Javier Pereira, researcher, Universidad Diego Portales

Samuel Sierra, student, Universidad Diego Portales

Sebastían Vera, student, Universidad Diego Portales

Brazil

Marcelo de Almeida Maia (local coordinator), researcher, Federal University of Uberlândia
Raquel Fialho Lafetá, PhD student, Federal University of Uberlândia

Fernanda Madeiral Delfim, PhD student, Federal University of Uberlândia

Eduardo Cunha Campos, student, Federal University of Uberlândia

France

Damien Cassou (local coordinator), associate professor, University of  Lille & Lifl/CNRS & Inria

Martin Monperrus, associate professor, University of  Lille & Lifl/CNRS & Inria

Romain Rouvoy, associate professor, University of  Lille & Lifl/CNRS & Inria

Jifen Xuan, researcher, Lifl/CNRS & Inria

Gérard Palligot, engineer, Lifl/CNRS

Martin Dias, PhD student, Lifl/CNRS & Inria

Gustavo Santos, PhD student, Lifl/CNRS & Inria

Matias Martinez, PhD student, Lifl/CNRS & Inria

Vincenzo Musco, PhD student, Lifl/CNRS & Inria

Benoit Cornu, PhD student, Lifl/CNRS & Inria

Maria Gomez Lacruz, PhD student, Lifl/CNRS & Inria


Gallery

First Workshop held in Federal University of Uberlândia in February, 2015.



Free Global Counter Flag Counter
« prev  |   top  |   next »

Published paper at JBCS Springer Open

Redocumenting APIs with crowd knowledge: a coverage analysis based on question types) with Fernanda, Klérisson, Damien and Marcelo published at Journal of the Brazilian Computer Society (Dec.2016).

Google Research Award Latin America

The PhD thesis proposal "Automated construction of tutorials from Q&A sites" from Adriano Rocha has been awarded in August 03, 2016. 

Published paper at Journal of Software: Evolution and Process - Wiley

The paper "Searching Crowd Knowledge to Recommend Solutions for API Usage Tasks" with Eduardo Campos, Lucas Souza and Marcelo Maia has been published in October 2016.

Powered by CMSimple | Template by CMSimple! | Login