Many with partial match consumes to much of the input about superpower

[nblumhardt]

Thanks for the report, @SeeDee82 - this is indeed a bug. I've created #78 to fix, including your test case.

I'll merge it through to dev so that a pre-release package is published; it would be fantastic if you can check that the new package behaves as you expect.

Thanks again!

[SeeDee82]

Thank you for the fast fix, this looks good.

On the subject of backtracking, can you explain why you decided to have backtracking only as opt-in. It seems to me that you would always want to have backtracking on by default.
The only two places I could find where backtracking is used at all are the Many and Or combinators, and in both cases I think you would expect partial failure and need the ability to try the next parser regardless of the reason for or the internal state after the failure.

[nblumhardt]

👍

Sprache (the inspiration for Superpower) does backtrack by default; in that approach you need to opt out of backtracking in order to improve error messages, rather than opt-in to make the parser work. Tests usually cover having the parser work, so optimizing for good error reporting by default means that the harder-to-test part (the error messages for scenarios the parser author might not have considered) is generally okay. HTH!