Disallowing "r#" at the beginning of identifiers

[garrisonhh]

As far as I can tell from the spec, r##### is a perfectly valid identifier. At the moment, given a sequence of characters representing a single kdl value or identifer, all types of values can be identified solely based upon their first two characters except for raw strings because of this particular case. For consistency, I'd also be in favor of disallowing # altogether from identifiers.

This is a tiny nitpick, but it makes sense to me from an implementer perspective!

[hkolbeck]

This definitely requires some attention. As much as it pains me, I'm in favor of disallowing # in bare identifiers. Currently kdl4j will reject bare identifiers starting with "r#".

From an impl perspective, allowing such identifiers requires unbounded lookahead as far as I can figure, which is certainly prohibitive. Because kdl4j is hand written I can certainly tweak things to allow it, but others implementations may not be able.

[zkat]

I don't think it's a tiny nitpick. I think it's pretty significant, and warrants having a discussion with implementers about putting out a new spec.

[zkat]

Can someone volunteer to put up a PR that disallows #? And maybe we can start a discussion about disallowing a couple more symbols, because I need some for KQL? :)

[hkolbeck]

I can do that. I'd like to have the discussion about other symbols to be disallowed sooner rather than later if possible, so I can do them all at once.

[esoterra]

Is that a WIP KDL-based query language?
If so, you'll have to compete with Kusto Query Language for SEO

[hkolbeck]

@Kylebrown9 It is, the spec is here: https://github.com/kdl-org/kdl/blob/main/QUERY-SPEC.md

[esoterra]

Oh, I see. It's a way of querying KDL documents nodes. That sounds quite useful for some applications.
For some reason I was expecting it to be a way to express database queries in KDL.

[pingiun]

I believe this is also the case with numbers. Consider the token +1214124 is this a number or an identifier? If we read the next character we can see it can only be an identifier +12124123k however if the next character were to be a space, this should've been read as a number instead. I think disallowing + and - at the start of identifiers would be a good fix for this.

[larsgw]

This was actually implemented already in #99 and then amended in #100 to allow for identifiers like --foo.

[pingiun]

Thanks for pointing me to this. I don't think it's explained very well in the prose and that's why I got confused. I'll take a look if I may be able to improve that wording

[larsgw]

Sounds good.