Unist vs Loyc trees about unist

[wooorm]

Cool project! Thanks for reaching out.

From what I gather on the site, it looks like Loyc is one format of syntax tree, for all programming languages. You still need parsers (document > syntax tree) and compilers (syntax tree > document) for every language. But there’s one syntax tree. Correct?

Unist is a standard on how to describe syntax trees. It provides a few essentials, but allows differences. The markdown syntax tree is different from the HTML syntax tree.

Right now, there’s lots of different syntax tree formats out there, and each format needs their own set of utilities to manipulate and inspect. Because of Unist, there’s no need to create those utilities for each syntax tree, just once: for Unist.

[qwertie]

You still need parsers (document > syntax tree) and compilers (syntax tree > document) for every language. But there’s one syntax tree. Correct?

Sorry for taking awhile to respond. What you mean by 'document > syntax tree'?

Basically, yes, there is one syntax tree, and moreover there is a single preferred syntax (which will be LESv3 when it is completed). Any other programming language can, in principle, be encoded as a Loyc tree, but Enhanced C# is currently the only other language with a Loyc tree representation.

The rest is open-ended, and I haven't decided what to tackle next. My original idea was a single compiler framework that allows multiple input syntaxes (e.g. imagine mixing Java, Swift and C# code in one program). Another possibility is a system that converts code from one programming language to another. Unfortunately both of these ideas require a lot more manpower. (for one thing, a multi-language standard library seems to be needed for both of them.)

A seemingly easier idea is a WebAssembly interoperability framework (ABI) that uses LES as a common language for cross-language bindings. (The challenge there is my lack of experience with building low-level tools, and the fact that WebAssembly isn't standardized yet.)

What programming language are you using for the code in the readme?

interface Node {
    type: string;
    data: Data?;
    position: Location?;
}

I guess this means "optional Data" and "optional Location"? So whereas a Loyc tree is "an identifier | literal | call node, plus a location and a list of attributes", A Unist is "a text | parent node, plus a location and some arbitrary Data and a type string". It seems like Unist is oriented around document storage, whereas Loyc trees are oriented around code. Yes?

[qwertie]

When it comes to document storage, I wonder, what's the advantage of Unist over a JSON encoding? (EDIT: sorry for misspelling Unist.)

[wooorm]

What you mean by 'document > syntax tree'

Here I define “document” as a file or fragment of a file. And “syntax tree” as a representation easily inspected by, and manipulated by, computers.

What programming language are you using for the code in the readme?

That’s WebIDL.

I guess this means "optional Data" and "optional Location"?

Yup!

It seems like Unist is oriented around document storage, whereas Loyc trees are oriented around code. Yes?

Unist and Loyc do different things, yes, but I’d say it differently: Unist is the interface of the tree, the format of the tree, what is expected at what keys, and what those keys mean. Loyc trees are one syntax tree for all programming languages.

Something else to consider is hast, and the like, which define a tree for one one programming language, based on Unist.

When it comes to document storage, I wonder, what's the advantage of Unist over a JSON encoding?

Unist, when represented in JavaScript, is just (parsed) JSON, yup! Probably in different languages too.
Think of Unist more as the schema and semantics.

---

Another possibility is a system that converts code from one programming language to another

That’s sort-of possible with this system too. Sure, the syntax trees are different, like hast for HTML and mdast for markdown, but remark-rehype can translate between the two trees: from markdown to HTML.

[wooorm]

@qwertie Did that help explain stuff? Can I close this?

[qwertie]

Yeah, I guess so. Some details are fuzzy to me, but my main question is answered, i.e., our projects are focused on different things.