DeepSpec is a software project funded by the National Science Foundation (NSF) in the amount of ten million dollars and carried by the Massachusetts Institute of Technology, the University of Pennsylvania and Yale University. The project goal is to develop an integrated system of tools to eliminate bugs in complex software systems. The project wants to be very ambitious. In fact, researchers around the project say their goal beyond basic research that is reshaping the industry itself in order to unite the progress that various researchers have had to make in this area. One goal is especially train the next generation of developers and engineers with good practice that could lead them towards the goal of zero bug in their programs.
The major challenge for the researchers, according to their team members, will come to understand and to master the complexity of modern hardware and software and identify important factors involved when two software components are put together to make a defined task. Another challenge, not least, will then make an accurate description of the behavior of software based on formal logic (deductive reasoning, using syllogisms, mathematics). This will eventually allow engineers to not only develop software without bugs, but also to check the conformity of these to believe the project members.
Such a project requires a lot of financial resources and a sufficiently long time to be completed. This was including the NSF by agreeing to finance it. Indeed, Jim Kurose who is the head of the foundation stated that: DeepSpec the program “allows the IT research community to continue the study of complex problems in computer over a long period.” He added that “this will enable these researchers to make progress not only in the disciplines of computer science, but also for other areas.
According to the research team, this project is timely, in a context where most of the technicians in the software industry consider writing software more like an art than a scientific discipline . This aspect is particularly programmers, working mostly on isolated tasks, do not care enough to document and codify their work so that others can use it as a base in their learning. Researchers on the project believe that the weakness of the institutional basis of knowledge has slowed progress in the search for solutions to what can be described as an enigma that is the resulting unpredictable behavior of putting together several programs to make given job. Andrew Appel, head of the research team, said in this sense: “Even if an engineer writing a software component and documents the English example, another engineer, he is an Englishman, can interpret this documentation in the wrong way.
Call is known for having participated in CompCert project of the French Institute for Research in Computer. Their work at that time consisted in creating a compiler that can accurately translate a programming language into machine instructions that can run on a smart card. According to Appel, the logical continuation of the current work is to connect components such as compilers, operating systems, program analysis tools, architectures processors, ensuring that no bug does manage to sneak between them. This is one of the objectives of DeepSpec that will facilitate this integration by improving the way in which the specifications are written using formal logic according Call. Another researcher who participated in the project name CompCert Beringer said that the project of the French Institute for Research in Computer has demonstrated that it is possible to write robust software specifications for complex software industry. He added that this model was followed by other teams on isolated projects in the design of software components. However, he said, the strength of DeepSpec is to bring together the efforts of all participants around the same problematic.
The project is conducted with the participation of several big names including Stephanie and Steve Weirich Zdancewic, both professors at the Department of Computer Science at MIT, Adam Chlipala, associate professor of computer science at MIT, and Zhong Shao, professor of computer science at Yale University. The work of Appeals revolve around tools for reasoning about computer programs that revolve around Steve compilers. Adam meanwhile will work on computer chips and Zhong take care to set up an audited operating system, for Stephanie, his work will turn around programming languages that can be used by developers to write their programs. Another member of the Benjamin Pierce name him team will work on software testing tools that will be based on the specifications.
Source: princeton.edu
And you
What do you think of this project? Is it realistic?
No comments:
Post a Comment