Do we really need ==:==?

[DmytroMitin]

==:== seems not to be needed.

SingleGeneric seems not to need Repr and Aux. Repr is always Store.

It's important to define instances of macro-based type classes like

// whitebox
def materialize[In, Out]: TC.Aux[In, Out] = macro materializeImpl[In]

rather than

// whitebox
def materialize[In]: TC[In] = macro materializeImpl[In]

Otherwise val tc = the[TC[A]]; implicitly[tc.Out =:= B] compiles but implicitly[TC.Aux[A, B]] doesn't. Similarly, otherwise db.users.store(user) compiles https://scastie.scala-lang.org/DmytroMitin/wEXiRryiSMaC1AiTYZVYZQ but db.users.store[User, User :: HNil, UUID :: String :: String :: HNil, User :: HNil](user) doesn't https://scastie.scala-lang.org/DmytroMitin/wEXiRryiSMaC1AiTYZVYZQ/5 . Ability to specify type parameters can be useful for example for debugging implicits.

It's a little weird that constraint implicit gen: Generic[T] is put for the type class SingleGeneric on method level rather than type-class instance level.

Maybe SingleGeneric can be ordinary type class (not macro-based)?

[alexflav23]

@DmytroMitin Long follow up, I am no longer actively managing the project and haven't been for quite a while. This makes a lot of sense, and it looks like it's working.

I am not even remembering the context, I recall there was some kind of special equality necessary to compare HLists, but not the specifics. Entirely possible I got something wrong.

Any way I can help integrate and publish these changes?

[DmytroMitin]

@alexflav23 Feel free to publish any changes you consider relevant. Although I'm not sure whether those changes are binary compatible (MiMa). CI seems to be broken.