kebab-case is less usable than snake_case

[mitranim]

I have usability complaints about kebab identifiers.

First off:

• I'm on board with the idea that the compiler should enforce a consistent identifier style. I just happen to think snake_case is the only good style.
• I understand that kebab-case is easier to type on most layouts. My custom layout has _ in place of =, so typing snake_case doesn't require Shift. I recommend that to everyone.

Now for the ergonomic annoyance. Excerpt from the code:

constraint : kw['constraint'] name-constraint ;

name-constraint{entity.name.constraint} : ident-any ;

create-domain : kw['domain'] name-type kw['as'] storage-type constraint? ;

alter-table-action-add :
  NONE
  | (kw['constraint'] if-not-exists? name-constraint (kw['not'] kw['valid'])?)
  | (kw['column']?    if-not-exists? declare-column)
;

alter-table-action-rename :
  NONE
  | (kw['to']         name-table)
  | (kw['constraint'] ident-other kw['to'] name-constraint)
  | (kw['column']?    ident-other kw['to'] name-column)
;

• Place cursor on the first constraint.
• Hit slurp_find_string (⌘E for me).
• Spam find_next (⌘G for me).
• Expectation: finds only exact matches in \b\b (muscle memory from other syntaxes).
• Reality: loads of false positives like name-constraint because according to \b\b, hyphen is a word delimiter.

Addendum: this particular case is solvable by using dots in names: name.constraint. In this case, that's a better naming scheme anyway. But this is not viable in all cases, and the dot may end up repurposed for namespacing when imports are added. The compiler doesn't let me do that.

[mitranim]

This also annoys me in YAML syntaxes. It's too late to fix that in YAML, but not too late for SBNF.

[BenjaminSchaaf]

loads of false positives like name-constraint because according to \b\b, hyphen is a word delimiter.

That's an issue in the core of ST, see: https://github.com/sublimehq/sublime_text/issues/4901. I'm going to close this as kebab-case is the accepted standard for syntax definitions (which are affected by the same issue) and it's clearly easier to write.

[mitranim]

I agree this should be fixed in the core to support syntaxes already committed to kebab-case. But I don't think it fully absolves kebab-case. Users may want to use regex search with \b\b for refactoring. Admittedly that's an edge case, but it shows that this word break problem "leaks" into many places.

[BenjaminSchaaf]

The core fix would make all search regexes use the correct word boundaries.