Robert Kruszewski

Currently at 80 bytes, mostly because it stores the dtype

DType enum is a bit too big

1

Currently at 48 bytes, could go down to 16 reasonably

Is it worth performing two passes to build a sorted dictionary to encode values with? There's likely cases where having sorted dictionary proves beneficial but not sure of tradeoffs

- [x] #10 - [x] #9 - [ ] #22 - [ ] #128 - [ ] #919

- [x] Projection pushdown - Closed by #568 - [ ] Filter pushdown - [ ] Take pushdown - [ ] Statistics pruning - [ ] https://github.com/spiraldb/vortex/issues/964

Enable https://rust-lang.github.io/rust-clippy/master/index.html#/unwrap_used and fix all the cases

Right now when creating Bytes based buffers we lose the alignment information. There has been no movement on issue to allow this https://github.com/tokio-rs/bytes/issues/437 which would mean that unless allocator_api is...