# Nushell core team meeting 2024-05-15
## Attendees
- Jack
- Ed
- Michael
- Ian
- Stefan
- Antoine
- Darren
- Sophia
- Devyn

## Agenda
- [x] [ByteStream PR](https://github.com/nushell/nushell/pull/12774), if anything to discuss, or questions
- [x] `Value::Table` / `PipelineData::TableStream`
- [ ] Config dirs, and especially distro-configurable ones
- [ ] ...

## Discussed Topics

## `ByteStream`

- should land soon, small feedback round with Devyn (thrown under the bus in absentia)
- let's see how much follow on work we can land before the release freeze
- (line ending stuff)

## Kicking `nu-cmd-dataframe`?

- Max-uvarov has been bouncing bugs with Jack back and forth
- [Announced plan for removal](https://hackmd.io/As6eZuAkRKC5trBoyFWoEQ)
- polars plugin starts to diverge in the removal of the eager mode behavior in favor of lazy
- (after the bytestream monster PR lands): remove the crate, feature `dataframe`, documentation, ci steps, build artifacts for `-full`

## plugin distribution
- Currently pretty `cargo` dependent
- `nupm` has plugins in its registry but runs `cargo` via build script
- future vision: responsible binary distribution

## internal table repr
- `Value::Table` and `PipelineData::TableStream`
- optimized form, still permit list of records
- could elide checks and column names

## cell paths and their pitfalls
- Ed: support for ranges in cell paths
- Stefan: only one row is permitted in a cell path (commutativity of row/column for table, but limitation for nested structures)
- Sophia: int vs string as the proxy for columns(record keys) and rows(list index)