
NPUC 2009 @ IBM Research - Almaden: Future of Design & Software Develepment: Rastislav Bodik, UC Berkeley
5 months ago
Abstract: Massive compute power has long been available to us but we need more ideas on how to harness it in programming. While testing and verification dutifully number-crunch programs, they do not fix the bugs they find. Moving closer to programmers, recent tools have become true cognitive assistants: search engines find relevant code samples, verifiers explain bugs, and software miners discover properties absent in the documentation. Still, these tools do not directly address the problem of writing programs.
This talk will describe a growing family of programmer tools that assist in writing programs. Their premise is that programs can be decomposed into insight and mechanics, and that the later can be synthesized from the former. The first research problem is how to describe the insight without spelling out the details. I will describe the idea of sketching (programs with holes) and how we move beyond programming by demonstration by asking an oracle for a demonstration. The second problem is how to synthesize the program from the insight. Some of our tools combine mining of examples with program analysis, and some rely on recent advances in decision procedures, such as SAT, that combine search with algebraic reasoning. I will conclude with open problems motivated by some of our more ambitious ideas.
About the Speaker
Ras Bodik is Associate Professor of Computer Science at UC Berkeley. He is interested in programming systems, from the HCI aspect of programmer tool and language design, to program analysis, compilation, and computer architecture. He has worked on mining of program specifications, debugging. He is currently leading projects on synthesis for programmer masses and on a web browser for mobile devices.
About NPUC 2009:
Software development and design has evolved from an arcane art practiced with exotic, obscure tools into a multi-billion dollar industry based on even more exotic and obscure tools. Given the advances that have been made in user interface technology and design processes, shouldn't we be able to create a better user experience for design and development?
At NPUC 2009, IBM Research - Almaden brought together innovators in academia and industry who are leading the way to a more natural, accessible, and social design and software development process . They explored how to help improve the productivity of current programmers and make creating software more accessible to a larger and more diverse population than ever before.
This talk will describe a growing family of programmer tools that assist in writing programs. Their premise is that programs can be decomposed into insight and mechanics, and that the later can be synthesized from the former. The first research problem is how to describe the insight without spelling out the details. I will describe the idea of sketching (programs with holes) and how we move beyond programming by demonstration by asking an oracle for a demonstration. The second problem is how to synthesize the program from the insight. Some of our tools combine mining of examples with program analysis, and some rely on recent advances in decision procedures, such as SAT, that combine search with algebraic reasoning. I will conclude with open problems motivated by some of our more ambitious ideas.
About the Speaker
Ras Bodik is Associate Professor of Computer Science at UC Berkeley. He is interested in programming systems, from the HCI aspect of programmer tool and language design, to program analysis, compilation, and computer architecture. He has worked on mining of program specifications, debugging. He is currently leading projects on synthesis for programmer masses and on a web browser for mobile devices.
About NPUC 2009:
Software development and design has evolved from an arcane art practiced with exotic, obscure tools into a multi-billion dollar industry based on even more exotic and obscure tools. Given the advances that have been made in user interface technology and design processes, shouldn't we be able to create a better user experience for design and development?
At NPUC 2009, IBM Research - Almaden brought together innovators in academia and industry who are leading the way to a more natural, accessible, and social design and software development process . They explored how to help improve the productivity of current programmers and make creating software more accessible to a larger and more diverse population than ever before.
-
Vimeo: About / Blog / Developers / Jobs / Community Guidelines / Community Forums / Help Center / Site Map / Merchandise
/ Get Vimeo

Previous Week