Inline test sources in mill - with hacky demo project...

[Quafadas]

I was reading / thinking about this thread on scala contributors which wants in-source testing / evaluation.

Figured I'd have a go at it in mill for the lolz. Getting "something" seems to be far easier than I would have imagined. Here's an example project.

In line test - https://github.com/Quafadas/fmt_rep/blob/932d55086286f816d8617b3b84093845db6107ea/submod/src/addit.scala#L6

And it seems to "work"... https://github.com/Quafadas/fmt_rep/actions/runs/20665359528/job/59336359516

In that it runs that test on CI. When I inspected the de-compiled compilation output, tests were not present in the submod compile output, but were present in the inlineTests module. So I believe the "compilation concept" works as far as it goes (duplicates the work, but it's CPU work).

Here's the (possibly bonkers) stuff I did in that build to make it work;

• Two mill modules, both source generating a testing function...
  • one generates a function which inlines it's contents - inline def apply[A](a: A) = if(true) a
  • The other generates a function which... doesn't - inline def apply[A](a: A) = if(false) a exterminates the code as far as I can tell.
• Give inlineTests module the same sources as the standard module.
• Add the testing framework
  • to mvnDeps in the testing module
  • to compileMvnDeps in the "publish" module

marks out of 10?

To be honest I'm surprised / happy that it seemed relatively "easy" to get something.

Are there any obvious reasons this is a terrible idea :-)? Anyone interested in exploring futher?