Calculating type sets is harder than you think
Go 1.18 added the biggest and probably one of the most requested features of all time to the language: Generics. If you want a comprehensive introduction to the topic, there are many out there and I would personally recommend this talk I gave at the Frankfurt Gopher Meetup.
Implementation restriction: A compiler need not report an error if an operand’s type is a type parameter with an empty type set.
This constraint can never be satisfied. It says that a type has to be both the predeclared type int and have a method M(). But predeclared types in Go do not have any methods. So there is no type satisfying C and its type set is empty. The compiler accepts it just fine, though. That is what this clause from the spec is about.
This decision might seem strange to you. After all, if a type set is empty, it would be very helpful to report that to the user. They obviously made a mistake - an empty type set can never be used as a constraint. A function using it could never be instantiated.
I want to explain why that sentence is there and also go into a couple of related design decisions of the generics design. I’m trying to be expansive in my explanation, which means that you should not need any special knowledge to understand it. It also means, some of the information might be boring to you - feel free to skip the corresponding sections.
