Not excited about this feature, I guess we'll see how frequently it shows up in unwanted places. Generics in C++ really damage the readability of the code sometimes, maybe the go devs have found a better way. Very skeptical, but the go devs have given me plenty of pleasant surprises before, maybe we get another one here.
I did a quick look through all my open source projects to see where I have accepted or returned an interface{} in my own APIs, which is a strong signal that I'm looking for generics, or am using a library that wants generics.
One case is interacting with Kubernetes. I have two applications where I set up a reflector like `cache.NewReflector(lw, &v1.Node{}, store, resync)`, to keep a cache up to date; the implementation of "store" takes interface{} for all the arguments (Add, Delete, etc.), even though it only ever operates on a v1.Node. Generics will let me ensure at compile time that I only have to worry about v1.Node objects. Right now, that's handled with runtime assertions in the reflector itself, and in every implementation of the cache.Store. Generics clean this up.
Another case that comes up is processing random JSON objects from the Internet with no defined schema. Those end up as map[string]interface{}, and that won't change.
The last case is unmarshaling functions. I see some (in a JWT validation library) that are of the form `func Unmarshal(string) interface{}`. I don't really know what's going on there, and I don't know if generics will help. (Similarly, for the common case of unmarshaling JSON, I'm not sure generics can improve the `err := json.Unmarshal(bytes, &result)` API. I will have to look into it.
That's about it.