r? @GuillaumeGomez

(rust_highfive has picked a reviewer for you, use r? to override)

r? @steveklabnik

I'm okay adding this, but i've noticed the reference and the grammar disagree with respect to how they define idents. The reference has additional rules stating that identifiers may be prefixed with an underscore, despite underscores not having XID_start. If you want to update the reference to take out its "_ alone is not an identifier" remark, I would suggest adding the rules about idents prefixed with an underscore to grammar.md. Since this is a pre-existing issue, I won't gate the PR on it, but if you want to do that follow-up you mentioned in your initial comment, that's what I would suggest. The reference is in a separate repo, so two PRs will be necessary.

@bors r+ rollup

📌 Commit 723772f has been approved by QuietMisdreavus

@ubsan
Could you talk about the motivation a bit?
I'd say _ is perceived more like punctuation intuitively and it's also is not treated as a keyword internally (the difference is mostly philosophical though).

@petrochenkov correct, it is mostly philosophical. I just thought this was logical, and simpler than treating _ specially

ident_or_kw := <start> <continue>*
start := <XID_start> | '_'
continue := <XID_continue>