Frank Thomas

Of course it must be hexagonal! :-) I like that toy @soronpo. A stylized version of that cube rotated such that it fits into the dimensions of the other Typelevel...

Other ideas: * a [compass](https://en.wikipedia.org/wiki/Compass_(drawing_tool)) maybe with a capital `T` for type as a symbol for making types more precise * a capital `T` where the left half is in...

@kusamakura Nice! I very much like the idea of something that goes from pixelated to high-res.

@kusamakura That would be great, thanks. I would prefer SVG.

That's interesting! I've never worked with an explicit `Inference` instance in user code. The `auto.autoInfer` macro cannot work in this case since it needs to evaluate `B ==> A` at...

I tend to change the error message from `Predicate failed: (4 > 6)` to something like `Predicate failed: (x > 6) where x = 4`. That should help to keep...

The `Inference` instances that depend on other `Inference` instances are now duplicated. One version for `InferAlways` and one version for `InferWhen`. I would prefer if we could achieve the same...

Forget that, my last suggestion does not work for ```scala implicit def hypotheticalSyllogismAlways[A, B, C](implicit p1: A ==> B, p2: B ==> C): A ==> C = Inference.combine(p1, p2, "hypotheticalSyllogism(%s,...

I've thought about this a bit more and I think I made a mistake in the design of `Inference` that we now have to pay the price for. The mistake...

The more I think about this issue, the more I become convinced that we should remove `isValid` from `Inference`. Pros for removing that field: * `==>` becomes similar to `=:=`...