The Pragmatic Embedded Podcast

Why Testing Sucks, Revisited

Episode Summary

Nearly 100 episodes after "Why Testing Sucks," Jeff and Luca revisit testing strategy with fresh perspectives. Jeff and Luca discuss when to write and remove tests, and structuring them to tell clear stories. We cover Test-Driven Development and when it helps versus hinders during prototyping. A major theme is how LLMs are changing testing. They ease refactoring and creating infrastructure like embedded simulators, but generate bloated suites full of mocks that may not give needed confidence. We emphasize keeping LLMs on a tight leash.

Episode Notes

Nearly 100 episodes after "Why Testing Sucks," Jeff and Luca revisit testing strategy with fresh perspectives.

The conversation centers on Guillermo Rauch's advice: "Write tests, not too many, mostly integration." We explore what this means, whether the traditional testing pyramid holds up, and what counts as "unit" versus "integration" tests—terms used loosely in practice.

Jeff and Luca discuss when to write and remove tests, and structuring them to tell clear stories. We cover Test-Driven Development and when it helps versus hinders during prototyping.

A major theme is how LLMs are changing testing. They ease refactoring and creating infrastructure like embedded simulators, but generate bloated suites full of mocks that may not give needed confidence. We emphasize keeping LLMs on a tight leash.

The episode covers requirements management, traceability, and how tooling encourages early test creation. The focus is practical heuristics: What are you worried about? What gives confidence? What story do your tests tell?

Key Topics

Notable Quotes

"The best testing heuristic you can use is, what am I actually worried about? Like, what do I need to guard against?" — Luca Ingianni

"Any complex working system was evolved from a simple working system. You need to start with Blinky of the LED and then add... you build it up from those simple beginnings." — Jeff Gable

"Not writing tests because you have considered them and you've considered them a waste of time is very distinct from not writing tests because you're in a hurry. That is a very dangerous trap to fall into." — Luca Ingianni

Resources Mentioned