James Clark

If we do `\b`, we should certainly do `\B`. This isn't in the JSON schema interoperable subset: https://datatracker.ietf.org/doc/html/draft-bhutton-json-schema-01#name-regular-expressions. Nor is it in I-Regexp https://datatracker.ietf.org/doc/draft-ietf-jsonpath-iregexp/ (which is what JSON schema is...

Positive and negative lookahead (but not lookbehind) are also in the JSON schema regex subset specified here: https://json-schema.org/understanding-json-schema/reference/regular_expressions.html

Looking at the notes on "look behind" here https://www.regular-expressions.info/lookaround.html , I think "look behind" is significantly more problematic than "look ahead", so I think we should have separate issues for...

Need to think about how interpolation (`${...}`) works. Currently it is defined to take values of the interpolated data type, but I think we also need it to takes strings....

A bigger issue would be to tie this into resource functions somehow (but resource paths aren't URLs).

I agree: capturing should be treated as an assignment (except when the variable is final or could have been final).

I also wonder what happens when the same variable is captured multiple times with different narrowings in effect for each occurrence.

We need to allow the error type of a named worker to be inferred #45. We originally had `error` for this but there were some bad interactions with `distinct`, so...

We need to way to say at compile time that a named worker does not outlive its function. This is useful when the function containing the workers is isolated: it...

I agree this would be useful.