Macropipe backlog¶
Iteration +1¶
Allow registering functions under different names / aliases.
Unlock using “regular” Polars expressions directly on columns, using a special syntax like
col=foo:expr:arg1,arg2.Add sorting for standalone use. When importing into a database, you don’t necessarily need it, but of course in many other cases.
Optimize pipe littering across the board.
Map more primitives: fill_null, drop_nans, drop_nulls, fill_nan.
Manipulation of nested data (slicing, reformatting), using json_path_match.
Iteration +2¶
Todo
- Q:
Can primitives functions like
headortailthat barely need any argument processing be mapped directly?- A:
Well, because of the macro nature, at least all types must be casted appropriately from
str.headandtailare excellent examples where only a single function argument needs to be processed (casted). Maybe we can invent some automatic mapping, so we don’t need to enumerate each and every sensible Polars method manually?
Use
responderfor a basic example to provide a HTTP interface to a Polars dataframe.How to sample random records?
Map or wrap recipes from polars-url, polars-ts, turtle-island, see awesome-polars.
Iteration +3¶
Insights into working with
try_parse_dates.https://docs.pola.rs/user-guide/transformations/time-series/filter/
What about multi-stage pipelines?
https://docs.pola.rs/polars-cloud/integrations/airflow/#parallel-query-execution
What about resampling?
https://docs.pola.rs/user-guide/transformations/time-series/resampling/
Timezone manipulation?
https://docs.pola.rs/user-guide/transformations/time-series/timezones/