# Nushell core team meeting 2023-04-05

## Attendees

- Stefan
- Antoine
- JT
- Michael
- Jakub
- Fernando
- WindSoilder

## Agenda

- First impressions after the new release
- https://github.com/nushell/nushell/pull/8635#issuecomment-1497535256 mentions a downside of the eager number parsing in non-command positions
- IDE support
- https://github.com/jntrnr/nushell-lsp VSCode extension
- Path forward for the VSCode extension
- https://github.com/jntrnr/nushell/tree/ide contains the modifications to `nu` (Landed! https://github.com/nushell/nushell/pull/8745)
- support for multiple static errors
- What other features do we want?
- Untested: multi-file support
- Skipped: Error types in general: [Taxonomy of errors in nushell](/RzY-4hEdSlO1LxE0e-amqw)
- Stuff to ship after the release
- std library/prelude
- collection of fixes and minor breaking changes
- Passing of flags/named args
- No more coverage?

# Discussed Topics

## Public feedback 0.78.0:
### number parsing

We now eagerly assume that code starting with a number is likely a number and respond with errors
- Semver:
https://github.com/nushell/nushell/pull/8635#issuecomment-1497535256
- IPv4 addresses

Let's wait and see. We had a conflicting flood of issues that favor having better number literal error messages.

## Update on the arrow development
arrow/arrow2 and polars seem to get closer so might be a better foundation when we can get Fernando back on the dfs

## IDE support

Currently a separate VScode extension that links up to new options on the `nu` binary

- type hints
- basic goto def
- hover with info/command docs
- error reporting (with one error)

Still alpha stage

global variables only
Jakub: scope dropping happens makes it more complex `parser_info`

still TODO link it up with the main vscode extension

### looking for performance issues with the interactive use

Inlay type hints seem to be slow accross the board and might be part of VS code's implementation

Darren: jakt uses some throttling mechanism -> worth looking into

## `match` command

Most constant forms should be covered
several PRs added it. (documentation of all the cool features)

## 1.0 outlook

we are just 22 releases away from the spiritual 1.0 :P

Feedback from Fernando: virtual environment handling with different language tools still a hassle (e.g. `poetry` )

What is our answer to "we want posix"/"we want bash compatibility"?

"just add one more eval, one more eval bro"

## STD library

- the problem we discussed last meeting seems to be resolved
- https://github.com/nushell/nushell/pull/8627
- `parse_module` to load the baked std library
- fixes the mutable wizardry
- is prelude ready (i.e. no need for `use`)
- example commands right now: assert commands
- `help` is not yet part of std.
- https://github.com/nushell/nushell/pull/8505
- `$nu` still misses a few fields
- command type and category not exactly
- signature works in individual help but not in the listing yet
- works in repl/scripts now
- *SHIP IT!!!*
- Darren: coordination around help/prelude + removal of rust commands
- make nu `help` implementation in prelude `std help` at first!

## "kwargs"/ passing a flexible set of flags on top of rest args

passing flags is helpful for wrappers/ externs is highly desirable

technically difficult to get typechecking/keeping positioning/and get flexibility

pythons `**kwargs` are a hellhole to loose documentation of arguments

we need to deal with the complexity as we want to be a shell

Powershell for reference
https://learn.microsoft.com/en-us/powershell/module/microsoft.powershell.core/about/about_functions_advanced_parameters?view=powershell-7.3

can we fall back to stringly world in this case safely

## PR's

### https://github.com/nushell/nushell/pull/8610

cell path on `$in`
currently only `update`

closure arg and `$in` would be different (is this confusing)

Wind: For insert, does it make sense to follow cell path on $in? Because the column we want to insert is not exists in $in, the same thing happened on upsert command.

General feeling try it on `update` -> land and try