Russell Mull

> I think that would only work for the rowan-like design, where concatenating the leaf nodes' text returns the original source. Coming back to the OrderByExpr example, it represents the...

This is /exactly/ the use case I had in mind for #161.

Rationale: Cache behavior can differ between ARM implementations, including qemu. It's possible to make changes which appear to be unrelated to this but acutally end up triggering a cache coherency...

> Without extern "C", does Rust follow these rules? Or does it always do the pointer-into-the-stack method? Whatever it happens to do, it's intentionally undefined, so we can't rely on...

See also https://crates.io/crates/cargo-call-stack

I've just run into the backtracking problem myself. I think the issue is that it's natural to expect the combinators to behave opaquely; having one branch of a choice fail...

Yes, this is clear. I'm suggesting that the invariants could be checked with a debug assert (or perhaps some other mechanism), to save future users some debugging time.

This looks very useful, it would certainly clean up some of my parsers. @cenodis would you consider releasing this as a separate crate, if it doesn't look like it's making...

In case anybody else needs it: I turned the contents of this PR into a little crate that works with nom 7: https://github.com/mullr/nom-7-precedence. I'm looking forward to being able to...

Here's the workaround I'm using for now. It's a bit of a cudgel, but afaict it gets the old behavior back. ```rust /// This is derived from nom::number::complete::float, which is...