Jeroen van Straten

When I was hacking away at my own parser stuff (before giving up and trying Chumsky instead), my approach for recovery for alternations was: - for each alternative: - parse...

Ah, > Because chumsky is a PEG parser, which always take the first successful parsing route through a grammar, recovering from an error may cause the parser to erroneously miss...

The primary reason I gave up writing my own library is because I confused myself with all the type/template stuff involved with combinators, but I'll give it another shot now...

Okay, went a little bit nuts and basically just started from scratch. Notably, I handled source location tracking a bit more generically, because I found myself fighting Chumsky a bit...

As of yet it's not completely done and usable yet; I realized earlier this week that I had forgotten to actually implement `recovers_with()` so there wasn't really a way to...

Got recovery combinator type derivation working by adding a random PhantomData; not sure why that made it work exactly, but having previously refactored my entire codebase three or so times...

Okay, I think I'm pretty much done now. I ran into a bunch of ownership issues when trying to implement something half-serious, so it took me longer than I hoped...

> Chumsky works fine on tokens that contain references. However, the main use-case for zero-copy parsing is the ability to refer back to long tracts of the input stream (i.e:...

Makes sense. It definitely seems like a "don't touch if you don't know what you're doing" thing, and I understand the maintenance aspects of exposing internals like that. It's a...

I don't feel qualified to comment on those design questions, but FWIW, I ran the serialized output of the test case through the validator and it's okay-ish (the validator doesn't...