internal: Try adding a label to a chumsky parser

[max-sixty]

Moving this from #4700. Do we know why adding a label removes the error message? Is it something to do with merging multiple labels loses the labels themselves

prql/prqlc/prqlc-parser/src/error/parse_error.rs

Lines 281 to 292 in 6357126

	impl SimpleLabel {
	fn merge(self, other: Self) -> Self {
	match (self, other) {
	(SimpleLabel::Some(a), SimpleLabel::Some(b)) if a == b => SimpleLabel::Some(a),
	(SimpleLabel::Some(_), SimpleLabel::Some(_)) => SimpleLabel::Multi,
	(SimpleLabel::Multi, _) => SimpleLabel::Multi,
	(_, SimpleLabel::Multi) => SimpleLabel::Multi,
	(SimpleLabel::None, x) => x,
	(x, SimpleLabel::None) => x,
	}
	}
	}

?

Is this the correct design? I would think we want the most specific item, not sure whether that's possible.

(Tagging @m-span as he was the most recent author of the code, but no worries if you're not sure / or if this was the result of a refactor. And having new line isn't important per se, I mean it as a more general question)

[m-span]

Do we know why adding a label removes the error message? Is it something to do with merging multiple labels loses the labels themselves

That code is copied from the default Chumsky implementation. My best guess it is so that when an error propagates upwards, only the first, most specific error-label is kept.

[m-span]

Still, I'll look into this case

[max-sixty]

That code is copied from the default Chumsky implementation. My best guess it is so that when an error propagates upwards, only the first, most specific error-label is kept.

Ah great, I see, it's explained here: zesterer/chumsky#238

So it seems like we just need to add labels to the "higher" item so that's reported, rather than nothing. (I need to think through what counts as "higher"...)

Still, I'll look into this case

Sure, if you want to take a look, but no obligation