Why You Need a Software Delivery Machine
summary
Claim: conventional YAML+bash pipelines break at scale — shown by a 958-line GitLab CI YAML example, hundreds/thousands of repos, a Canadian bank’s 47-slide delivery doc and a 100-page security rule set that never fully gets applied. Built an event-driven system using TypeScript workers, WebSockets and GraphQL that runs secret-scanning push listeners, reports to Slack, issues auto-fix commits and enforces goals. The approach works in production and lets org-wide policies replace duplicated per-repo pipelines.
AI-written summary · about
chapters (12)
- 0:00Intro
- 1:09My own conclusions...
- 7:43Function composition
- 9:41The Deeper Problem: Limited, Anemic Model
- 12:30Why This Matters: Practical Problems
- 20:10We're Setting the Bar Too Low
- 28:50Deploying apps was hard
- 32:50Software Defined Delivery
- 34:30Atomist's Implementation of SDD
- 39:08Case Study
- 40:39Characteristics
- 46:19Portable Programming Model
related talks
21mBuild your JS Pipeline in Incremental Fashion
31mSoftware Delivery and the Rube Goldberg Machine: What Is the Problem We Are Trying to Solve?
46mThe Software Factory
36mCI/CD BoF: Delivering Quality Quickly
55mContinuous Deployment Workflows for Modern JavaScript
45mBecoming a Plumber: Building Deployment Pipelines
40mSecure by Default: Building an AI-Augmented, OSS-Powered Reusable CI/CD Pipeline
59mDevOps for the Discouraged