For the last 13 years at Microsoft, I have dived into many code bases trying to understand complex concurrency patterns. Generally, this has been amazing experience, but sometimes I see patterns of usage that my theory brain just does not want to see. Typically, after much deliberation, I come to conclusion this is okay. In this talk, I will present some of these examples that could potentially be used as challenge problems for academia, or at least will make the audience squirm.

Matthew Parkinson is a principal researcher at Microsoft in Cambridge, and is actively involved in the development of Project Verona and snmalloc. His research is focused around memory and concurrency safety. He has published papers ranging from formal verification to systems and language design.