Hyeseong Kim

I'm building a code generator, so I just wanna refer an existing type by its name. Is it not the common case?

It does only guarantee the output structure is the same, and it does not point to the same type(name). Do I understand correctly?

The record definition spread is useful for forcing not to reuse value type in ReScript. But when interacting with structurally typed language, I often want the opposite. In the example...

The word "nullish" has never been adopted by the ecosystem. I think that would be a little strange. The words I see the most are "maybe" or "optional".

It is called as nullable for even TypeScript people: https://www.typescriptlang.org/docs/handbook/utility-types.html#nonnullabletype Maybe we can remove something inconsistent, but I don't think the "nullable" is not a problem

I'm not actively promoting this to avoid adding any more noise before the v11 release. And still having time to think about better refactoring and migration strategies :grinning: But please...

> Do we want to keep dev-dependencies? Is there a clear use-case for it? no use cases right now unless we plan to support a standalone package manager for `rescript.json`....

No. It also happens in v9.x. Maybe it was there from the beginning of the new syntax

@DZakh noticed [another pattern](https://twitter.com/dzakh_dev/status/1609945222879870976) related to this The props mapping is sometimes used even within higher-order functions. But this proposal does not handle the case, for example: ```res let fn...

~~Another issue is the genType doesn't reflect PPX's results at all. I think it should~~ ---- It can be done by PPX implementor, see #6537