Player

1 obviously doesn't work because you need to get your classes loaded by knot cl 2 is where you can't use your own ModCandidateFinder because it's package private and faking...

Letting mods control the mod set is quite an important feature, it allows all sorts of updaters, adapters, server-mod-synchronizers and even your application. I expect to land the feature set...

This area is still very WIP, I wouldn't spend too much time translating there. The overall presentation and content of these error texts isn't anywhere near it should be. I...

I don't really mind making the directories configurable this way for special purposes, the way it is configured already targets advanced users. Disabling the mods directory should be possible too,...

I don't quite understand how this is supposed to work, normally fabric loader or launchwrapper controls the program entry point. The proposed method implies it runs somewhere in between, but...

I think so, see https://openjdk.java.net/jeps/277 - "the API is experimental and is subject to incompatible changes"

I forgot to mention there that adding such a mechanism would imply doing the same for blocks. I don't really want to clutter the interface with a generic arg for...

Longer term I was considering to change the API such that the resolver object gets created with a src,dst namespace pair and drops the 1st arg in map*.

Not sure what effect this is supposed to have? There is the "provides" system which can (besides other things) expose shaded mods (currently not with its own version, but that'll...

I still don't get it? Please try to be more verbose. At least with the current capabilities of the dependencies block it makes little sense to me.