DESCRIPTION

This is a talk about the design strategies I employed to build a package that can tokenise JSON 8-10x faster than the standard library.In the talk I'll cover the high level design of the package. I'll show how design a API that avoids allocation and unnecessary copying (this is a major contributor to the performance of the library). Finally I'll touch on some of the optimisation decisions in the package and show how they work in sympathy with inlining and bounds check elimination.