Projects

• Programming by Sketching: Program synthesis that programmers can use. Project wiki.
• Parallel Web Browser: Browsers and languages for parallel handheld devices. Why browsers will be parallel (Slides).
• Program Analysis and Tools: pointer analysis, incrementalization, thin slicing.

Recent Presentations

• Browsing Web 3.0 on 3.0 Watts: why browsers will be parallel and implications for education. Keynote at STMCS 2008. This talk presents the vision behind our parallel web browser project and simultaneously makes a case for revising undergraduate education in programming languages. April 2008.
• Program Synthesis by Sketching. Plenary talk at P=AC² 2008. An overview of our program synthesis project. April 2008.
• Why do we still have bugs. A panel presentation on why we should be so lucky to always have many bugs. More than semi-serious. P=AC², March 2008.

Collaborations

• Par Lab is a collaborative playground for my group's parallel programming projects.
• Many of my software quality projects are done under Open Collaborative Research with IBM.
• The OSQ project is an umbrella for most PL research done at UC Berkeley.

Teaching. My students and I have been developing a new language/compiler undergraduate course (CS164) which I am teaching this semester.

Software. Software released by our group can be found here.

Students. Students I am currently working with: Gilad Arnold, Apollo Ellis, Chris Jones, David Mandelin, AJ Shankar, Armando Solar-Lezama, Manu Sridharan. The following students have graduated.

Research interests: program analysis, synthesis, and compilation, both static, dynamic, and hybrid; tools for software developers; hardware support for debugging, testing, and compilation.

Publications. You can find most of my papers here. Please my students' pages for related slides. Also, here is my cv, a bio, and list of awards.

Recent activity in my group: About half of our excitement is in sketching, our approach to bringing software synthesis to hands of regular programmers. There are currently two sketching projects: the SKETCH langauge and the Prospector synthesizer. Besides exploring software synthesis, the later project also looks at new ways to mine software. We also work on precise analyses for realistic software, most recently for pointer analysis. We are always interested in non-traditional compilation techniques: recently, we had very interesting results in partial evaluation and in incrementalization. We are also interested in software tools: most notably, we developed a new definition of slicing that appears to give much more usable slices.

• Programming by sketching: sketch an outline of your implementation and the compiler will complete it for you:
  • Sketching Stencils, PLDI'07 (more info)
  • Programming by Sketching for Bitstreaming Programs, PLDI'05 Best Paper Award (software, talks)
  • Combinatorial Sketching for Finite Programs ASPLOS'06 (more info)

• JavaSketch and Prospector: another sketching project, a programmer's search engine/code synthesizer based on jungloid mining
  • paper: Mining Jungloids: Helping to Navigate the API Jungle, PLDI'05
  • JavaSketch download (at sourceforge)
    
    
• Program Analysis:
  • Refinement-Based Context-Sensitive Points-To Analysis for Java, PLDI'06
  • Demand-Driven Points-to Analysis for Java, OOPSLA'05
  • ABCD: Eliminating Array-Bounds Checks on Demand, PLDI'00
    
    
• Software Tools:
  • Thin Slicing (a poster), PLDI'07
  • Ditto: Automatic Incrementalization of Data Structure Invariant Checks, PLDI'07

• Hardware support for debugging, dynamic analysis and optimization:
  • Runtime Specialization With Optimistic Heap Analysis, OOPSLA'05
  • Regulated Transitive Reduction for Longer Memory Race Recording, ASPLOS'06 (a short article in IEEE Micro)
  • Serializability Violation Detector for Shared-Memory Server Programs, PLDI'05
  • A "Flight Data Recorder" for Enabling Full-system Multiprocessor Deterministic Replay, ISCA’03
  • Rapid profiling via stratified sampling, ISCA'01
  • An Efficient Profile-Analysis Framework for Data Layout Optimizations, POPL'02

Past Projects

• BAFL: Bottleneck Analysis of Fine-grain Parallelism
  • Focusing Processor Policies via Critical-Path Prediction [ISCA'01]
  • Slack: Maximizing Performance under Technological Constraints [ISCA'02]
  • Using Interaction Costs for Microarchitectural Bottleneck Analysis [MICRO'03]
  • Software release: Critical-path analysis and criticality prediction for simplescalar (for now, please ask for it).
    
    
• Specification Mining: Learning Specifications of Software Correctness
  • Mining Specifications [POPL'02]
  • Debugging Temporal Specifications with Concept Analysis  [PLDI'03]
  • this project has evolved into the jungloid mining project called Prospector (see above)
    
    
• Pathfinder: path-sensitive value-flow optimizations
  • annotated bibliography

Misc

• OSQ Retreats: 2005, 2006, 2007
• Berkeley PS Seminar
• Wild and Crazy Ideas at ASPLOS 2002.
• OM 2001: ACM SIGPLAN Workshop on Optimizations of Middleware and Distributed Systems.
• sumbit a paper to GPCE