Scaling Full-Stack Applications
summary
A production Nubank back‑office called Shuffle person deck runs ~75k lines of ClojureScript, 15,000 commits, 120 contributors and deploys 3–4× daily; the team builds a declarative data layer using petal resolvers (Pathom/EQL style) plus Fulcro, Datomic and spec to express attribute relationships, automatic orchestration, caching and backtracking. Resolvers declare inputs/outputs so a single customer ID drives composed UI queries, tracer and workspace tools expose request timing and eliminate duplicated requests; demos show YouTube, SpaceX and GraphQL adapters working end‑to‑end.
AI-written summary · about
chapters (32)
- 0:00Introduction
- 0:33Shuffle
- 1:05Project Numbers
- 1:28Where is my data
- 3:32Problems with the architecture
- 4:41Select keys
- 6:46Demo
- 17:50Request Caching
- 21:10Petal
- 21:43User Interface Visualization
- 22:36React Formula
- 24:17Focus
- 26:16Compose Components
- 28:01Workspaces
- 28:39Basic Keywords
- 29:08FookRespect
- 29:36Transactions
- 29:58Network
- 30:17Exploratory
- 30:44Tests
- 31:01Lazy Properties
- 31:37Incremental Loads
- 32:30Monitoring
- 32:47Testing
- 33:10Browser Demo
- 39:26Composition is Key
- 40:19Future Projects
- 40:29Improvements
- 41:03Graph to Graph Communication
- 41:29GraphQL API
- 41:48GraphQL Provider
- 42:15Thank you
related talks
38m"Growing Data Center networking mgmt UI using ClojureScript, Reagent and re-frame"
39mWeb Apps in Clojure and ClojureScript with Pedestal
31mClojure at 4,000 msg/s-What We Learned, Loved, and Loathed
26mFrontend At Scale - Experiences Using React In A Large Organization
33mLightning talks
29mDesign decisions in React-based ClojureScript applications
27mThe Functional Final Frontier
30mBringing it all together