`nufmt` roadmap

[AucaCoyan]

We have the repo in the nushell org 🍾 . Now we need more structure to keep moving forward. As we work in issues and features, we can edit the list.

⚠️ restructure the project

Because of the new-nu-parser coming, and the issue with DeclIds we need to restructure a lot of the parser as it is today. I don't know how long this is going to be, but we have to have some patience. Let me investigate until I can plan what to do and when 📆

❗ TODO

• ⚠ resolve all the FlatShapes in #13 (see above)
  • [x] #35
  • [ ] there are still some missing shapes
• [ ] with the file formatted as (--minified), we need to add the indentation
• [ ] make configurable with Config struct
• [ ] limit the max characters to Config
• [ ] and endings of ;, if they are relevant to nushell logic (delete them if they are unnecessary, like python)
• [ ] option to write the files in place or output another file
• [ ] Format reports are nice. Add them!
• [ ] public the cargo package
• [ ] add the nixpkg auto-publish workflow
• [ ] Make it available as a GitHub Action
• [ ] add some ignore comment (# nufmt ignore)

✅ Done:

• [x] make clippy happy
  • #12
• [x] accept to format a string, not just files.
  • #15
• [x] clean /examples/ folder and move them into tests. Add benchmarks
  • #16
• [x] adjust README.md
  • [x] add the supported nu version
  • [x] add badges
  • #16
• [x] #17
  • [x] #18
• [x] check if the last line is a new line and, if false, add one.
  • #28
• [x] examine the input path: Is it a file? format it. Is it a directory? format the .nu files and skip the other files
  • #47

[ItsHarper]

Given the slow progress of nufmt, does it make sense to adopt topiary-nushell as nufmt for now? Once it's ready, this codebase could replace it down the line (similar to the plan for https://github.com/nushell/new-nu-parser), as using the real nu engine is undoubtedly more powerful.

See also https://github.com/blindFS/topiary-nushell/issues/37

[fdncred]

I recommend topiary-nushell frequently because it's so good right now. However, I'm not sure I'd adopt it as the official one because it does not use the nushell parser. It uses tree-sitter which nushell does not use. So, there's a chance these two parsers could/do diverge. I'm sure there are already things that are implemented in either parser that the other doesn't really support right now.

What we need is more contributors to nufmt.

[blindFS]

What we need is more contributors to nufmt.

Agreed, but new-nu-parser is of higher priority IMHO.

[fdncred]

Agreed, but new-nu-parser is of higher priority IMHO.

Ya, you're probably right since nushell's parser isn't lossless. All it takes is a dedicated few to finish it and start implementing it.