DIFFY Catching Bugs without Writing Tests
summary
Diffy is an open-source tool from Twitter that detects regressions in backend services by automatically comparing responses from candidate and baseline deployments as part of continuous deployment workflows. The talk explains Diffy’s core concepts and how its automation integrates into CI/CD to reduce manual testing effort for service owners, DevOps, and QA.
AI-written summary · about
chapters (14)
- 0:00Intro
- 1:48The Problem
- 2:43The Other Problem
- 3:08Unit Tests
- 4:04Component Tests
- 4:54Integration Tests
- 6:19emerging pattern ...
- 6:46Diffy Approach - more coverage for free
- 8:16but ... what about noise?
- 9:44Diffy Topology
- 20:12Automation
- 23:44Performance Regression Detection
- 24:38Diffy-Performance Topology
- 27:49Composed Classifiers
related talks
38mDiffy - Predicting and Preempting Regressions in Services - Puneet Khanduri & Prashant Khanduri
22mDiffy: Data-Driven Bug Finding for Configurations
34mCan your diff(1) do this?! Improving software review and QA with diffoscope
26mServer-Side Diff: Future-Proofing ArgoCD's Comparison Engine
28mWhat the Diff?
59mDevOps for the Discouraged
39mYou're Testing WHAT?
16mDiffStream: Differential Output Testing for Stream Processing Programs