Gabriel Scherer

Minor point: instead of having a zoo of weird non-variance signs that we always never see in code and are not self-discoverable (good luck googling for `^'a`), I think that...

(also @let-def obviously, as the author of the related work on CAMLroots)

> First: performance aside, could the boxroot API be implemented using just a doubly linked list for bookkeeping? Yes -- we would use a generational design two linked lists, one...

If I understand correctly, the proposal is to use a non-generational allocator (create new boxroots, either old or new, in a single container (pool, doubly-linked list, etc.), and in addition...

@stedolan this sounds similar to our implementation, but a bit different. (I think our "pools" are your "chunks", we have "minor" pools that contain a mix of young and old...

Note: we discussed this RFC again at our last development meeting, and the consensus was in favor of us submitting a PR to consider upstreaming a Boxroot implementation. Note that...

I must admit that I didn't push for this RFC in any particular way. Florian, if you are willing to openly declare yourself as a supporter, maybe we should start...

The advantage of the `%%unit_header` proposal is that it requires no syntactic change. This said, I think the proposal of using a new level for unit-global annotation is also fine...

Apparently a [similar Rust RFC](https://rust-lang.github.io/rfcs/3503-frontmatter.html) has just been approved. Their main use-case seems to be single-file programs (`cargo script`), rather than module-specific data in a larger program.

I had a brief discussion about this with @Octachron a few days ago, and we agree that accepting a preamble of toplevel directives would be nice. This would push us...