Is TDD dead? [Part III] (feedback)

https://www.youtube.com/watch?v=YNw4baDz6WA

===

Kent

Feedback

instant feedback is not always possible

how far can we?

1) how frequently feedback?

100ms -- minutes

2) fidelity of feedback

green + prod -> % no crash

different kinds of software

different personalities

3) overhead

4) lifespan

years & probabilities

===

Martin

there is more than one thing you want feedback on

1) User needs

"is this software doing something useful for the user?"

"how do I want to visualise the UI"

2) Have I broken anything?

3) Is the codebase healthy?

===

David

QA people were cut off because TDD was over-succesful

over-confidence in tests thought up by you during development

test-induced damage

tradeoff needs to know the costs

but TDD only has more tests, faster

what are you giving up, is it worth giving up?

===

Kent

QA, swung too far

dysfunctions of QA

throw over wall

long cycles

big feedback

responsibility

"nothing at facebook is someone else's problem"

===

Martin

thoughtworks QA

===

David

We're wilfully trading off some degree of quality for now, because we wanna move fast, break things and may not be around tomorrow

new generation

shallow understanding of what quality is

programmers who did not want to be on call

===

Kent

The on-call is a feedback loop about test you did not write