Prior thinking

[wooorm]

Hey this looks interesting! Nice work.

It’s not really the same, but I’ve spent the last five years thinking about similar problems, with unist, mdast (markdown), hast (html), and nlcst (natural language).

Feel free to close, just wanted to share this!

[earthlyreason]

RDF is a thing. JSON-LD is a thing. Since the OP and the author have undergone the immense efforts to make these things with the obvious aim of avoiding reinvention, please consider the addition of a globally-unique key in which the semantics informally defined here could be made unambiguous.

[wooorm]

@gavinpc-mindgrub I’m sorry, I don’t get what you’re suggesting?

[earthlyreason]

@wooorm, sorry I was not clear. From what I can tell, this project and the ones referenced by the OP all define ad-hoc schemas for representing marked-up text as JSON. They also appear to intend these schemas for use outside of their particular tools (though I could be wrong about that). The "official" method of supporting interop like that would be to define an RDF vocabulary to give each key an assigned meaning in a given context. To some extent this is just about a commitment to the schema as it stands, along with more formal documentation at an explicit location. JSON-LD makes it about as frictionless as possible to do this using a @context key. This context can be implicit within the tool itself (where you know that the schema is expected), but would allow annotation of data on the wire with an unambiguous signal of what it represents.

Saw this on HN and just wanted to mention it. I understand if it's not considered in scope.