Test architecture is a real thing
summary
Argues that test architecture matters by showing concrete choices: end-to-end tests on Neat’s embedded devices, a python-behave page object model that encapsulates GUI selectors (namedtuple + ditify) and computes shortest navigation paths, a retry framework (“henr”) that records micro‑failures, Elastic/Kibana trends that surface regressions instead of single noisy runs, a Slack “Autobot” to run targeted suites, and an auto‑triage with RandomForest model retrained daily using confidence thresholds (autoscrubs at 60% or 50pp gap). Each measure reduces false positives, speeds fixes, and—by the author’s report—works in production.
AI-written summary · about
chapters (10)
- 0:00Test architecture concept
- 5:54Defining architecture
- 15:36Artifacts and page object model
- 26:52GUI references and navigation
- 33:45Avoiding test automation blind spots
- 39:52Chaos in embedded testing
- 43:00People-centric tools
- 45:10Auto-triage with ML
- 50:11Environmental limits and sustainability
- 53:01Closing remarks and Q&A
related talks
58mTesting Software Architecture
51mTesting Software Architecture
44mTesting Software Architecture
40mTesting Software Architecture
43mTesting Software Architectures
20mEffective Testing Approaches for your Application with Cecelia Martinez
51mA testable architecture
01h 34mThe Intersection of Architecture and Implementation