Repository navigation

Issues

New issue

Tracking issue for the new streaming engine
#20947 · coastalwhite opened on Jan 28, 2025
95
Tracking issue for Polars Cloud
#21487 · ritchie46 opened on Feb 27, 2025

Nested rolling in rolling/group_by_dynamic aggregations returns wrong values (in-memory engine)
pola-rs/polars#30328 · opened on Oct 26, 2026
Wrong max/n_unique after aggregating a literal column with empty groups
pola-rs/polars#30326 · opened on Oct 26, 2026
sort() of a per-group scalar in group_by().agg() or over() reads out of bounds (segfault)
pola-rs/polars#30325 · opened on Oct 26, 2026
Add sort function to selectors
pola-rs/polars#30324 · opened on Oct 26, 2026
GPU engine rejects dt.offset_by("1d").dt.truncate("5m")
pola-rs/polars#30323 · opened on Oct 26, 2026
GPU duplicate sort keys with opposite directions fail after a projection
pola-rs/polars#30322 · opened on Oct 26, 2026
Integer literals larger than Float16 max overflow to inf in Float16 arithmetic, even when the result fits
pola-rs/polars#30319 · opened on Oct 26, 2026
1
Parquet writer fails when value sizes are skewed, leaving file half-written
pola-rs/polars#30312 · opened on Oct 26, 2026
write_json stops at the first empty chunk, writing [] or a truncated array
pola-rs/polars#30309 · opened on Oct 26, 2026
Corrupt file after sink continues writing to previously failed sink
pola-rs/polars#30305 · opened on Oct 26, 2026
2
SQL: IS [NOT] TRUE and IS [NOT] FALSE return null for a null operand
pola-rs/polars#30303 · opened on Oct 25, 2026
Dot-path access to nested struct fields, e.g. pl.path("a.b.c")
pola-rs/polars#30301 · opened on Oct 25, 2026
Projection pushdown into Parquet struct fields (read only the leaf columns a query needs)
pola-rs/polars#30300 · opened on Oct 25, 2026
NodeTraverser.view_expression loses nanosecond precision for Datetime literals
pola-rs/polars#30284 · opened on Oct 24, 2026
rolling_* error message still says min_periods instead of min_samples
pola-rs/polars#30265 · opened on Oct 23, 2026
Add a skip_blank_lines option to read_csv / scan_csv
pola-rs/polars#30254 · opened on Oct 23, 2026
[Regression] read_csv with new_columns shorter than file width raises SchemaError in v2.0 (worked in v1.x)
pola-rs/polars#30249 · opened on Oct 23, 2026
2
LazyFrame.describe() / with_columns(pl.all().sort()) uses ~37× data size in memory on the streaming engine
pola-rs/polars#30238 · opened on Oct 22, 2026
1
SQL: CROSS JOIN UNNEST(col, col) with the same column twice panics
pola-rs/polars#30208 · opened on Oct 22, 2026
1
scan_delta with a DeltaTable source reads the latest table version after pickling or serialize()
pola-rs/polars#30205 · opened on Oct 22, 2026
Add Arrow support for new mssql-python driver in read_database
pola-rs/polars#30195 · opened on Oct 22, 2026
GPU rejects dt.offset_by("40m") with a Temporal function 24 diagnostic
pola-rs/polars#30193 · opened on Oct 22, 2026
GPU rejects map_elements(math.erfc) with an anonymous function diagnostic
pola-rs/polars#30189 · opened on Oct 22, 2026
GPU rejects a column expression as the start offset to str.slice
pola-rs/polars#30187 · opened on Oct 22, 2026
Support scalar subqeries in nested join subquery (SQL)
pola-rs/polars#30183 · opened on Oct 22, 2026

Navigation Menu

Repository navigation

Issues

Issues

Nested `rolling` in `rolling`/`group_by_dynamic` aggregations returns wrong values (in-memory engine)

Wrong `max`/`n_unique` after aggregating a literal column with empty groups

`sort()` of a per-group scalar in `group_by().agg()` or `over()` reads out of bounds (segfault)

Add `sort` function to `selectors`

GPU engine rejects `dt.offset_by("1d").dt.truncate("5m")`

GPU duplicate sort keys with opposite directions fail after a projection

Integer literals larger than `Float16` max overflow to `inf` in `Float16` arithmetic, even when the result fits

Parquet writer fails when value sizes are skewed, leaving file half-written

`write_json` stops at the first empty chunk, writing `[]` or a truncated array

Corrupt file after sink continues writing to previously failed sink

SQL: `IS [NOT] TRUE` and `IS [NOT] FALSE` return null for a null operand

Dot-path access to nested struct fields, e.g. `pl.path("a.b.c")`

Projection pushdown into Parquet struct fields (read only the leaf columns a query needs)

NodeTraverser.view_expression loses nanosecond precision for Datetime literals

`rolling_*` error message still says `min_periods` instead of `min_samples`

Add a `skip_blank_lines` option to `read_csv` / `scan_csv`

[Regression] read_csv with new_columns shorter than file width raises SchemaError in v2.0 (worked in v1.x)

LazyFrame.describe() / with_columns(pl.all().sort()) uses ~37× data size in memory on the streaming engine

SQL: `CROSS JOIN UNNEST(col, col)` with the same column twice panics

`scan_delta` with a `DeltaTable` source reads the latest table version after pickling or `serialize()`

Add Arrow support for new mssql-python driver in read_database

GPU rejects `dt.offset_by("40m")` with a Temporal function 24 diagnostic

GPU rejects `map_elements`(`math.erfc`) with an anonymous function diagnostic

GPU rejects a column expression as the start offset to `str.slice`

Support scalar subqeries in nested join subquery (SQL)