Error Recovery

[DavidSouther]

As I understand it, today, Ohm's parser will stop on the first syntax error. What would it look like to have error recovery available and implemented? I see a couple issues and discussions, namely

#387
#302
#265

[pdubroy]

I definitely find the labeled failures approach interesting! I'd be happy to support you (or anyone else) who's interested in digging in to try to prototype this.

[DavidSouther]

I dug at it a couple years ago, but I can't find my notes from the time. I dug at it again more recently, and came up with ! as a "Fallible" marker, simplifying the Labeled Failure approach a bit. The idea being A B! C would allow B to fail with an error. In practice, I could see a language using this like statement = expression ! ';' - a statement tries to read an expression, but if it can't, it "resets" at the next ;. This essentially works as a non-greedy * and captures that information into an "Error" semantic type.

I don't remember where this ended up in practice, but here's the diff: DavidSouther#1

(I also have this change to streamline testing using the editor, it looks like - DavidSouther/ohm-editor#1)

I might have some time in the next couple weeks to pick this up again, but no promises!