You don't need a full-featured testing framework, or something that outputs in JUnit format, to get the value of a red-green-refactor loop.
All you need is a feedback mechanism, which can be as simple as "does the exit code for this process indicate an error or success?".
Kent Beck's test-commit-revert demo using python does exactly this - no importing a framework or test runner, but leaning on python's built-in `assert` keyword.
https://youtu.be/Aof0F9DvTFg?t=220&si=D7afjzT2GvTEhTQn
A guided example through what the different type of annotation ElementTypes are, with respect to Bean Validation, and how to write unit tests for them.
The thing that I found most valuable - and most generally underrated - from learning TDD 5 years into my career is the art of developing by making lots of small, targeted commits that each deliver an increment but leave the codebase deployable. Multiple commits per day.
I describe TDD, and practically every other software development technique as "neither necessary nor sufficient, but sometimes useful".
Zealotry of TDD or any other practice is often a consequence of Dunning-Kruger; a lack of awareness of how effective other techniques can be.
IMO, a snazzy rebrand at this stage would only be helpful if we could also stop people framing TDD as a silver bullet, and acknowledge that it is just one of many useful strategies that have different Value vs. Cost trade offs in different circumstances.
Seems unlikely though..🙂
I hear some "OGs" telling devs you don't need TDD anymore after you've done it "long enough"
The design feedback and safety and 20 other things TDD gives us, even the oldest practitioners we know we can't do without. I personally will never be ditching it.
Don't buy it.
#TDD
Moving into Quality Engineering, publishing Chef training courses, conference speaking about Chef at OggCamp and the complex mess that is this very static website at DevOpsDays London.
Jamie Tanna
.
#tdd.
