David Tolnay

Triage: I still think a resumable/nonblocking way to deserialize specifically a serde_json::Value is the way to go; we don't need to solve the much bigger challenge of resuming an arbitrary...

I would prefer not to have this in this library, but it would be reasonable for someone else to implement a more fully featured YAML library which implements this stuff.

This seems relevant: https://github.com/Keats/validator. Let's make sure to support this library in a nice way.

#626 has another use case. I am on board with implementing exactly the approach in @erickt's comment.

We are brainstorming some ways to generalize this in #1118.

Thanks for the update! Right now I don't have a great sense of what the people relying on this would prefer to optimize for (whether it's performance, binary size, etc)....

This is a bug -- I would accept a PR to fix this.

Wow, thanks. Can you share the code for your wrapper or send a PR?

Hmm, that's unfortunate all around. We cannot implement Drop for Value because it breaks existing destructuring code, we cannot implement Drop for Vec\ because it overlaps Vec\'s impl, we cannot...

Thanks for reporting this! That is definitely substantially longer than expected. Looks like practically all of that time is spent in LLVM: ```rust $ cargo rustc --release -- -Z time-passes...