As::Query Design Notes

[librasteve]

As the implementation of DataFrame and Series contexts and transformations matures, there is a potential to combine the slightly different APIs of Python-Pandas and Rust-Polars and unify them with the initial implementation of Raku-Dan. So the Dan API v1.0 will change to the Dan API v1.1.

This post seeks to compare and contrast the various API alternatives and to nail down the unified interface definition for this new API version.

1. Dan API v1.0

• As::Array is the default API, with:
  • 2d data delegation
  • [cascading] accessors
  • elems
  • shape
  • splice (deprecate, see below)
• As::Query
  • concat
  • sort
  • filter [grep]

novel (vs. raku Array)

2. Dan::Polars API v1.1

• As::Array is same as Dan
• As::Exprs is new (col prerequisite for As::Query, or spoof)
• As::Query is based on Polars examples, with following

Method	Example	Notes
series	df.series()	[~column]
select	df.select([col("sepal.length"), col("variety")])	:(Expr @cols)
with_columns	df.with_columns([col("variety").alias("newnew")])	:(Expr @cols) [~add]
drop	df.drop(["sepal.length", "variety"])	:(Str @cols) [tbd]
groupby	df.groupby(["variety"]).agg([col("petal.length").sum])	:(Str @cols)
agg	^^^	:(Expr @cols)
filter		[aka grep]
sort
concat	dfa.concat( dfc )	adapted Python(Pandas|Polars) [~concat]
join
=> !methods
column	df.column("sepal.length")	[~series]
with_column	df.with_column($se.rename("newcol"))	special case of add/with_columns [n]
collect

3. Next Action

• implement new drop, refactor, submethods to Dan::Polars
• move As::Query to role
• circle back to match in Dan & Dan::Pandas