On Rails — Aaron Patterson: Ractors, JIT Compilers, and Rewriting How Gems Install

Aaron Patterson, a senior staff engineer on Shopify's Ruby infrastructure team and contributor to Ruby and Rails Core, discussed on the On Rails podcast how Ractors are bringing true multi-core parallelism to Ruby and proposed content addressable gems to RubyGems.org to speed up bundle install. Patterson also shared his daily use of AI, including building a JPEG encoder almost entirely with Claude, while expressing lingering distrust of AI compared to human teammates.

In this episode of On Rails, Robby is joined by Aaron Patterson, better known online as Tenderlove, a senior staff engineer on Shopify’s Ruby infrastructure team and a longtime contributor to both Ruby and Rails Core. Aaron and Robby chat about how Ractors are finally bringing true multi-core parallelism to Ruby, the proposal Aaron has pitched to RubyGems.org for content addressable gems that could make bundle install dramatically faster, and the real reasons gems reach for native C extensions in the first place. They also have a candid conversation about how Aaron actually uses AI day to day, from digging through unfamiliar code at Shopify's scale to building a JPEG encoder almost entirely with Claude, and why he still can't quite bring himself to trust it the way he trusts a longtime teammate. Links TenderloveShopify Ractors RubyGems.org FFI