port build from sbt to mill? about coulomb HOT 8 CLOSED

Use the following as a basis for replacing sbt clean unidoc ghpagesPushSite
com-lihaoyi/mill#1194

from coulomb.

can use this: ScaladocSiteModule

com-lihaoyi/mill#1194 (comment)

from coulomb.

I believe I've demonstrated nearly all that I need here:
https://github.com/erikerlandson/mill-testbed/blob/main/build.sc

Except also should test scalaJS building

from coulomb.

I now have a working jvm/js cross-platform example, so I think I am in a position to port the coulomb built to mill.
Will probably do this by constructing the Scala-3 build from the ground up using mill, and leave the legacy scala-2 libraries using sbt.

from coulomb.

What's the benefit of switching to mill. I have no experience with mill as to tell what the benefits are

from coulomb.

@cquiroz I've always found sbt's config model to be confusing. Probably the most comprehensive answer to "why mill" or "why not sbt" was written by @lihaoyi :
https://www.lihaoyi.com/post/SowhatswrongwithSBT.html

In my proof-of-concept testing, I feel like I confirmed that mill is easier to reason about than sbt, although such things are always subjective:
https://github.com/erikerlandson/mill-testbed/blob/main/build.sc

I think that no build system can ever be truly "easy", mostly because build systems by necessity live at the intersection of a large number of kinds of functionality, which are discussed here:
https://www.lihaoyi.com/post/BuildToolsasPureFunctionalPrograms.html

So my current point of view is that while mill doesn't magically make everything super easy, it's a solid improvement over sbt. Again, though YMMV.

I guess I'll add link to his third blog post on mill
https://www.lihaoyi.com/post/MillBetterScalaBuilds.html

from coulomb.

Thanks for the detailed write up. I haven't really tried, as hard as it i to understand sbt I have (after a long, long time) managed to make my peace with it. Maybe I should revisit mill

from coulomb.

For now I'm feeling OK about using sbt with sbt-typelevel, so I'll close this

from coulomb.