core: Fix GitHub syntax highlighting

[danielbayley]

Force GitHub to highlight *.ys files as YAML.

[ingydotnet]

Awesome. I didn't know about this.

I can apply this for now but we should possibly look at other syntaxes.

Weirdly, highlighting as powershell works pretty good in vscode...

[danielbayley]

we should possibly look at other syntaxes.

Ideally, a YAMLScript: entry would be added to linguist here, the library responsible for syntax highlighting on here… But I think they have minimum constraints around popularity/usage, so the project will probably need to grow a lot more before they would accept a PR…

Weirdly, highlighting as powershell works pretty good in vscode...

PowerShell is also listed, so you could change this to the following, if that would render better highlighting…

*.ys linguist-language=powershell

[ingydotnet]

do you know if *.ys applies to code sections like

```ys
!yamlscript/v0
say: 42
```

[danielbayley]

do you know if *.ys applies to code sections like

```ys
!yamlscript/v0
say: 42
```

I think that would only work after a merged PR entry to linguist… Until then, would have to just do:

``` yaml
!yamlscript/v0
say: 42
```

or

~~~ yaml
!yamlscript/v0
say: 42
~~~

[ingydotnet]

I can certainly test it out on a branch,,,

[ingydotnet]

I realized that I can retarget the PR myself.
Retargeted to devel.

[ingydotnet]

@danielbayley do you have any experience creating a new syntax highlighter for linguist?

[ingydotnet]

Merged into a branch intended for 0.1.88

[ingydotnet]

YAML syntax highlighting is fine for now, but I'd like to get real ys token highlighting.

[danielbayley]

@danielbayley do you have any experience creating a new syntax highlighter for linguist?

@ingydotnet Iv'e done plenty of parsing stuff, but not really much in the context of syntax highlighting—apart from minor related things in Atom packages… I imagine Linguist probably uses something like TextMate grammars, or something RegEx-based.

I'd like to get real ys token highlighting.

Yeah ideally! I will say for sure, the much better approach would be the Tree-sitter/AST route…

[ingydotnet]

I've also done a bunch with parsing including creating the Pegex grammar language: https://github.com/pegex-parser
and also a little bit of syntax highlighting for CodeMirror, but not much.

I think the big thing that YS needs first is a language server.`
The YAMLScript compiler is perfectly postured to work as an LSP implementation for YAML and YS.

Here's a visual of the comilper in action:

$ ys -cde 'say: 2 + 3'
*** parse     *** 0.211807 ms

({:+ "+MAP", :! "yamlscript/v0/code"}
 {:+ "=VAL", := "say"}
 {:+ "=VAL", := "2 + 3"}
 {:+ "-MAP"}
 {:+ "-DOC"})

*** compose   *** 0.005074 ms

{:! "yamlscript/v0/code", :% [{:= "say"} {:= "2 + 3"}]}

*** resolve   *** 0.042651 ms

{:xmap [{:expr "say"} {:expr "2 + 3"}]}

*** build     *** 0.108232 ms

{:xmap [{:Sym say} {:Lst [{:Sym add+} {:Int 2} {:Int 3}]}]}

*** transform *** 0.027301 ms

{:xmap [{:Sym say} {:Lst [{:Sym add+} {:Int 2} {:Int 3}]}]}

*** construct *** 0.051574 ms

{:Top [{:Lst [{:Sym say} {:Lst [{:Sym add+} {:Int 2} {:Int 3}]}]}]}

*** print     *** 0.004313 ms

"(say (add+ 2 3))"

(say (add+ 2 3))

The output of the build stage should have all the info we need, except for input position (line/col).
The parser that YS uses does report that information but it is currently ignored.