My point was it's a top-down process with different layers of tests and POC.

The API tests of the top layer are stable, the POCs less so.

There is no point in starting a long endeavor without

• clarifying the clients expectations
• proving the feasibility

This is best done with some demo code, and this can and should be tested. Written specifications are open for interpretations, tests let you sleep well.

Since those tests define your API they can't easily be thrown away anymore unless your POCs were bad.

Hope my perspective is clearer now.

Yea again this assumes your demo code proves to be correct on the first swing, and I can count on one hand I've seen that happen.

I honestly don't see the point in writing tests for demo/poc code until you and the client are in agreement that the current stage looks correct. Changes after that are changes in scope and will probably require tests to be reworked anyway.

How do you know your top level API tests are stable? How do you know they're going to stay that way?

Three thousand years of beautiful tradition, from Moses to Sandy Koufax, you're god damn right I'm living in the fucking past

I think, we may have very different applications in mind.

> How do you know your top level API tests are stable?

Because they cement the specifications of the project.

The only problem I see with my approach, is that clients tend to consider demo code as production ready.

But I may have problems to express my ideas properly here.

Cheers Rolf
_{(addicted to the 𐍀𐌴𐍂𐌻 Programming Language :)
Wikisyntax for the Monastery}