True omniscient debugging is a lofty goal: record a program execution, efficiently reproduce any desired program state from the recording, visualize those states across time, explore connections between states (e.g. data flow), and make it all work for real developers of real applications. In fact, this has actually been achieved — subject to various constraints — and users report large productivity gains. The next grand challenge is to make this technology ubiquitous by relaxing those constraints. I will present an overview of the design space, the current state of the art, and what has to be done in the areas of hardware, operating systems and programming language implementations as well as debuggers to get closer to all developers having access to omniscient debugging at all times.

In 2001 Robert graduated with a PhD in computer science from Carnegie Mellon University, supervised by Daniel Jackson and Jeannette Wing. His thesis was on using polymorphic type inference to perform context-sensitive alias analysis for Java code.

From 2001 to 2004 he worked at IBM Research on dynamic program analysis tools, mainly using Java bytecode instrumentation. At the same time he moonlighted as a volunteer working on the Mozilla open source project, which became Firefox.

In 2005 he moved back to New Zealand to work full-time on Firefox for Novell and then Mozilla directly, which he carried on until 2016. In the last few years at Mozilla he led the design of rr, a practical record-and-replay debugger supporting reverse execution, used by many developers inside and outside Mozilla to the present day.

In 2016 he left Mozilla to co-found Pernosco, a startup working on cloud-based, omniscient debugging leveraging rr and new technologies.