Home

Project News

July 5th, 2012:
Cloud9 has been released at http://cloud9.epfl.ch/.

November 1st, 2011:
Stefan Bucur joins Google Zurich for an internship.

Oct 27th, 2011:
Our paper "Data Races vs. Data Race Bugs: Telling the Difference with Portend" will be presented on March 6, 2012 at ASPLOS.

June 1st, 2011: 
AutoSRS has been covered in media, in the context of the EcoCloud official launch. Find the article here.
High-quality tools for software reliability should be as easy and ubiquitous as webmail !

We are building cloud-based automated software-reliability services (AutoSRS), a new way to make advanced reliability techniques accessible and affordable to everyone in the world. Research-wise, we aim to make fully-automated testing techniques (which today are mostly usable only on toy programs) scale to real-sized software by harnessing massive clusters of commodity hardware. AutoSRS is compelling also from a social and business point of view: universal access to top-of-the-line tools could improve global software quality in much the same way webmail influenced people’s ability to communicate.We envision three types of AutoSRS:
  • An automated testing service that thoroughly tests code as developers write it, pointing out hidden bugs on the fly, similar to how automated spell checking works in modern text editors;
  • An automated debugging service that takes a program + bug report and walks the developer through how that particular code could have reached the particular buggy state described in the report; and
  • public certification & rating service, akin to Underwriters Labs, which independently and automatically assesses the reliability, safety, and security of both commercial and open-source software.

AutoSRS at Google

We welcome any Googler interested in participating in this effort; after all, this is a joint project between DSLAB and Google. Please check out the list 20% project ideas, and/or contact Prof. George Candea if you would like to engage in any of these projects or suggest other ideas.

We are grateful to Google for supporting this work through a generous Focused Research Award.