feat!: Use 4 spaces for indentation. #390
Conversation
|
| token_with_location(value(Token::Indentation, tag(" "))), | ||
| preceded(many0(tag(" ")), lex_token), | ||
| ), | ||
| (lex_indent, preceded(many0(tag(" ")), lex_token)), |
There was a problem hiding this comment.
Thinking we should remove the preceded(many0(tag(" ")) part, since we're being more strict about indentation being four spaces.
| (lex_indent, preceded(many0(tag(" ")), lex_token)), | |
| (lex_indent, lex_token), |
There was a problem hiding this comment.
Yeah, I was expecting to remove that too, but it becomes clear pretty quickly that the preceded parser in question is actually necessary to catch spaces between instructions tokens.
There was a problem hiding this comment.
Do you have a concrete example of this? Instructions should not be separated by spaces afaik
There was a problem hiding this comment.
Oh yeah, instructions was the wrong word there, I meant tokens. Having an issue with my environment, but I'll share an example of how removing the preceded parser causes an error.
There was a problem hiding this comment.
Oh, that makes more sense. In that case, it seems like separated_list0 or separated_list1 seems more correct
There was a problem hiding this comment.
Hmm, I could be missing my shot in nom golf, but I can't figure out how to make that combinator work for this case. It doesn't satisfy the trait bounds of the other wrapping combinators.
There was a problem hiding this comment.
I believe my intent was something like this, which does change the signature of the right-hand parser from TokenWithLocation to Vec<TokenWithLocation>, but also seems more correct?
| (lex_indent, preceded(many0(tag(" ")), lex_token)), | |
| (lex_indent, separated_list1(space1, lex_token)), |
There was a problem hiding this comment.
I suppose, in order to make this "more correct", newline parsing would have to be handled by lex_indent (since an indent is specifically a newline followed by four spaces). That seems like it would cause more code changes, since it seems like newlines are currently handled by lex_token?
closes #385
TODO