Eric Myhre

I wonder if we might want to hoist those Optional and Nullable wrappers somewhere broader. For example, the gogen package also emits an awful lot of boilerplate ["MaybeT"](https://github.com/ipld/go-ipld-prime/blob/67a06f12fc43892780a38f97ebbc5bd80c60f486/schema/gen/go/genpartsCommon.go#L18-L56) types right...

There's a https://github.com/ipfs/go-unixfsnode/ repo nowaday, which I believe has even more of these features.

#249 is solid evidence that there is real user confusion right now, which we could fix by this work.

An additional improvement we could make as part of this would be ensuring that the compiled selector type is a relatively closed one: a struct with unexported fields could have...

I've wondered if we'd be able to get away with not needing this, but it seems like the consensus is no, we need it :) So, okay. Agree `KeyIterator()` and...

Okay, I finally have an update on this: this actually comes from a design flaw, so the fix may take a while. The good news is I think we've figured...

(I don't think this is much of a breaking change, either way -- this is a design issue I remember fumbling around the first time I wrote this stuff, and...

I lean towards Option 1 for several reasons: - less code - the `basicnode` package is for untyped data; admit it (and double down on simplicity that this allows) -...

I'm thinking it might be nice to move some of the concepts to the `go-ipld-prime/linking` package -- `Storer`, `Loader`, `StoreCommitter`, and `LinkBuilder` all seem to make sense nudged over that...

For future quick reference -- the interfaces and types discussed above currently look like this (comments stripped for brevity): ```go type Link interface { Load(context.Context, LinkContext, NodeAssembler, Loader) error LinkBuilder()...