Explore OpenJDK's implementation at Google in this 44-minute talk by Jeremy Manson. Gain insights into Java development practices, deployment challenges, and static analysis techniques employed by Google. Discover why coders initially resist static analysis and how they come to appreciate its benefits. Learn about global changes, non-build breaking errors, and dynamic analysis approaches. Understand the limitations of traditional profiling methods and delve into garbage collection optimization strategies. Examine experimental stack allocation techniques and their potential impact on Java performance at Google.
Overview
Syllabus
Intro
Java at Google
Development at Google
One Slide on Deployment Issues
Digression 1: Static Analysis
Coders don't like static analysis!
Mechanics
Global changes
Coders do like static analysis!
Non-Build Breaking Errors
Dynamic Analysis
Why Profiling Doesn't Work
What about other profiling?
GC Eats Up My CPU
In short: Make the GC Better
Experimental Stack Allocation
Taught by
Java
