NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
crates.io, #2710 most downloaded on crates.io
IO related logic for the Polars DataFrame library
Latest stable
0.55.2
Last release 3 months ago
06 Aug 2026
Ships fairly regularly
a new release about every 2 months
Rarely documented
notes for 3 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
6 years old
103 releases · first in 2021
Drag across the ruler to compare, or tap a release to open it.
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Your coding agent can read these notes before it upgrades. Set up the MCP server →
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
This version includes quite a few breaking changes. We are preparing for the 1.0 release and aim to make the upgrade from 0.20 to 1.0 as smooth as pos…
This version includes quite a few breaking changes. We are preparing for the 1.0 release and aim to make the upgrade from 0.20 to 1.0 as smooth as possible. Therefore, we prioritized getting any breaking changes in now rather than with 1.0.
Check out the upgrade guide for help navigating the upgrade to this version.
Please bear with us while we continue to make Polars the best tool it can be!
Enum categorical data type which allows a fixed set of categories (#11822)read_parquet (#13044)replace expression on the Rust side (#13002)update signature (#12986)Expr.count to ignore null values by default (#12934)DataType objects to be instantiated (#12470)value_counts resulting column name from counts to count (#12506)join behavior with regard to nulls, add join_nulls parameter to keep existing behavior (#12840)Null when no data is present (#12807)lit behavior for list/tuple inputs (#12559)DataType.is_nested from property to classmethod (#12453)NaN ordering to make NaNs compare greater than any other float, and equal to themselves (#12721)write_database parameter if_exists to if_table_exists (#12783)Series methods (#13010)Series.head/tail to the expression engine (#12946)any/all_horizontal (#12976)truncate (#12965)select_seq for expression dispatch (#12962)rolling_median algorithm (#12704)DataFrame.iter_rows for smaller buffer sizes (#12804)Series from a list of NumPy arrays (#12785)str.contains_any and str.replace_many (Aho-Corasick algorithms) (#13073).aws folder (#13062)scan_parquet (#13060)read_parquet (#13044)Series methods (#13010)replace expression on the Rust side (#13002)inefficient map_* warning (#13039)hist (#13014)describe to use new count implementation (#12990)to_struct Series name consistent with the usual default Series name (empty string) (#12998)map_elements" warning message (#12978)end before start in date/time_range (#12964)update signature (#12986)Array data type repr (#12973)Null dtype (#12975)Expr.count to ignore null values by default (#12934)repr of Struct data type class (#12922)merge mode to write_delta and remove pyarrow to delta conversions (#12392)str.reverse (#12878)DataType objects to be instantiated (#12470)value_counts resulting column name from counts to count (#12506)std and var for Duration columns (#12865)join behavior with regard to nulls, add join_nulls parameter to keep existing behavior (#12840)write_database return (indicate the number of rows affected by the operation) (#12830)Decimal selector (#12852)UInt power (#10446)__repr__ implementation for Expr (#12770)JOIN and FROM (#12819)quantile(method="nearest") (#13058)datetime_range if starting on ambiguous datetime and earliest was specified (#13050)json_decode per max buffer length (#13029)00:00 time zone as UTC (#13034)align_frames and fix edge-case where the identical frame object appears more than once (#13007)ranges (#11900)sink_csv (#12991)read_database calls against cursors that only take positional args (#12967)truncate when truncating by multiple weeks (#12948)Err result (#12953)ambiguous parameter is not Utf8 (#12913)rolling_var/rolling_std numerical stability (#12909)min/max due to incorrect SIMD mask construction (#12908)to_numpy in the absence of pyarrow (#12888)Enum types (#12886)Expr.gather (which was still showing deprecated take) (#12864)Array dtype equality (#12853)nan_min/max incorrectly aggregating chunks with addition (#12848)collect_all functions (#12796)group_by (#12304)0.20 (#12844)describe calculation of min/max (#13027)count (#12960)group_by_dynamic (#12906)--no-cov flag for py3.12/ubuntu test workflow (vs implicit/omitted) (#12889)hash docstring (#12879)list.take (#12873)list.take is deprecated (#12867)pip install with dependencies (#12799)update docstring #12797Thank you to all our contributors for making this release possible! @MarcoGorelli, @Object905, @Yerachmiel-Feltzman, @alexander-beedie, @c-peters, @ion-elgreco, @jankislinger, @mcrumiller, @nameexhaustion, @oli-clive-griffin, @orlp, @rancomp, @ritchie46, @romanovacca, @stinodego and @xuestrange
join behavior with regard to null valuesPreviously, null values in the join key were considered a value like any other value. This meant that null values in the left frame would be joined with null values in the right frame. This is expensive and does not match default behavior in SQL.
Default behavior has now been changed to ignore null values in the join key. The previous behavior
can be retained by setting join_nulls=True.
Example
Before:
>>> df1 = pl.DataFrame({"a": [1, 2, None], "b": [4, 4, 4]})
>>> df2 = pl.DataFrame({"a": [None, 2, 3], "c": [5, 5, 5]})
>>> df1.join(df2, on="a", how="inner")
shape: (2, 3)
┌──────┬─────┬─────┐
│ a ┆ b ┆ c │
│ --- ┆ --- ┆ --- │
│ i64 ┆ i64 ┆ i64 │
╞══════╪═════╪═════╡
│ null ┆ 4 ┆ 5 │
│ 2 ┆ 4 ┆ 5 │
└──────┴─────┴─────┘
After:
>>> df1.join(df2, on="a", how="inner")
shape: (1, 3)
┌─────┬─────┬─────┐
│ a ┆ b ┆ c │
│ --- ┆ --- ┆ --- │
│ i64 ┆ i64 ┆ i64 │
╞═════╪═════╪═════╡
│ 2 ┆ 4 ┆ 5 │
└─────┴─────┴─────┘
>>> df1.join(df2, on="a", how="inner", nulls_equal=True) # Keeps previous behavior
shape: (2, 3)
┌──────┬─────┬─────┐
│ a ┆ b ┆ c │
│ --- ┆ --- ┆ --- │
│ i64 ┆ i64 ┆ i64 │
╞══════╪═════╪═════╡
│ null ┆ 4 ┆ 5 │
│ 2 ┆ 4 ┆ 5 │
└──────┴─────┴─────┘
Previously, the result of an outer join did not contain the join keys of the left and right frames. Rather, it contained a coalesced version of the left key and right key. This loses information and does not conform to default SQL behavior.
The behavior has been changed to include the original join keys. Name clashes are solved by
appending a suffix (_right by default) to the right join key name. The previous behavior can be
retained by setting how="outer_coalesce".
Example
Before:
>>> df1 = pl.DataFrame({"L1": ["a", "b", "c"], "L2": [1, 2, 3]})
>>> df2 = pl.DataFrame({"L1": ["a", "c", "d"], "R2": [7, 8, 9]})
>>> df1.join(df2, on="L1", how="outer")
shape: (4, 3)
┌─────┬──────┬──────┐
│ L1 ┆ L2 ┆ R2 │
│ --- ┆ --- ┆ --- │
│ str ┆ i64 ┆ i64 │
╞═════╪══════╪══════╡
│ a ┆ 1 ┆ 7 │
│ c ┆ 3 ┆ 8 │
│ d ┆ null ┆ 9 │
│ b ┆ 2 ┆ null │
└─────┴──────┴──────┘
After:
>>> df1.join(df2, on="L1", how="outer")
shape: (4, 4)
┌──────┬──────┬──────────┬──────┐
│ L1 ┆ L2 ┆ L1_right ┆ R2 │
│ --- ┆ --- ┆ --- ┆ --- │
│ str ┆ i64 ┆ str ┆ i64 │
╞══════╪══════╪══════════╪══════╡
│ a ┆ 1 ┆ a ┆ 7 │
│ b ┆ 2 ┆ null ┆ null │
│ c ┆ 3 ┆ c ┆ 8 │
│ null ┆ null ┆ d ┆ 9 │
└──────┴──────┴──────────┴──────┘
>>> df1.join(df2, on="a", how="outer_coalesce") # Keeps previous behavior
shape: (4, 3)
┌─────┬──────┬──────┐
│ L1 ┆ L2 ┆ R2 │
│ --- ┆ --- ┆ --- │
│ str ┆ i64 ┆ i64 │
╞═════╪══════╪══════╡
│ a ┆ 1 ┆ 7 │
│ c ┆ 3 ┆ 8 │
│ d ┆ null ┆ 9 │
│ b ┆ 2 ┆ null │
└─────┴──────┴──────┘
count now ignores null valuesThe count method for Expr and Series now ignores null values. Use len to get the count with
null values included.
Note that pl.count() and group_by(...).count() are unchanged. These count the number of rows in
the context, so nulls are not applicable in the same way.
This brings behavior more in line with the SQL standard, where COUNT(col) ignores null values but
COUNT(*) counts rows regardless of null values.
Example
Before:
>>> df = pl.DataFrame({"a": [1, 2, None]})
>>> df.select(pl.col("a").count())
shape: (1, 1)
┌─────┐
│ a │
│ --- │
│ u32 │
╞═════╡
│ 3 │
└─────┘
After:
>>> df.select(pl.col("a").count())
shape: (1, 1)
┌─────┐
│ a │
│ --- │
│ u32 │
╞═════╡
│ 2 │
└─────┘
>>> df.select(pl.col("a").len()) # Mirrors previous behavior
shape: (1, 1)
┌─────┐
│ a │
│ --- │
│ u32 │
╞═════╡
│ 3 │
└─────┘
NaN values are now considered equalFloating point NaN values were treated as unequal across Polars operations. This has been
corrected to better match user expectation and existing standards.
While this is considered a bug fix, it is included in this guide in order to draw attention to
possible impact on user workflows that may contain NaN values.
Example
Before:
>>> s = pl.Series([1.0, float("nan"), float("inf")])
>>> s == s
shape: (3,)
Series: '' [bool]
[
true
false
true
]
After:
>>> s == s
shape: (3,)
Series: '' [bool]
[
true
true
true
]
NaN equalityThe assertion utility functions assert_frame_equal and assert_series_equal would use the
tolerance parameters atol and rtol to do approximate checking, unless check_exact was set to
True. This could lead to some surprising behavior, as integers are generally thought of as exact
values. Integer values are now always checked exactly. To do inexact checking, convert to float
first.
Additionally, the nans_compare_equal parameter has been removed and NaN values are now always
considered equal, which was the previous default behavior. This parameter had previously been
deprecated but has been removed before the end of the standard deprecation period to facilitate the
change to NaN equality.
Example
Before:
>>> from polars.testing import assert_frame_equal
>>> df1 = pl.DataFrame({"id": [123456]})
>>> df2 = pl.DataFrame({"id": [123457]})
>>> assert_frame_equal(df1, df2) # Passes
After:
>>> assert_frame_equal(df1, df2)
...
AssertionError: DataFrames are different (value mismatch for column 'id')
[left]: [123456]
[right]: [123457]
DataType objects to be instantiatedPolars data types are subclasses of the DataType class. We had a 'hack' in place that
automatically converted data types instantiated without any arguments to the class, rather than
actually instantiating it. The idea was to allow specifying data types as Int64 rather than
Int64(), which is more succinct. However, this caused some unexpected behavior when working
directly with data type objects, especially as there was a discrepancy with data types like
Datetime which will be instantiated in many cases.
Going forward, instantiating a data type will always return an instance of that class. Both classes
and instances are handled by Polars, so the previous short syntax is still available. Methods that
return data types like Series.dtype and DataFrame.schema now always return instantiated data
types objects.
You may have to update some of your data type checks if you were not already using the equality
operator (==), as well as update some type hints.
Example
Before:
>>> s = pl.Series([1, 2, 3], dtype=pl.Int8)
>>> s.dtype == pl.Int8
True
>>> s.dtype is pl.Int8
True
>>> isinstance(s.dtype, pl.Int8)
False
After:
>>> s.dtype == pl.Int8
True
>>> s.dtype is pl.Int8
False
>>> isinstance(s.dtype, pl.Int8)
True
Decimal and Array data typesThe data types Decimal and Array have had their parameters switched around. The new constructors
should more closely match user expectations.
Example
Before:
>>> pl.Array(2, pl.Int16)
Array(Int16, 2)
>>> pl.Decimal(5, 10)
Decimal(precision=10, scale=5)
After:
>>> pl.Array(pl.Int16, 2)
Array(Int16, width=2)
>>> pl.Decimal(10, 5)
Decimal(precision=10, scale=5)
DataType.is_nested changed from a property to a class methodThis is a minor change, but a very important one to properly update. Failure to update accordingly
may result in faulty logic, as Python will evaluate the method to True. For example,
if dtype.is_nested will now evaluate to True regardless of the data type, because it returns the
method, which Python considers truthy.
Example
Before:
>>> pl.List(pl.Int8).is_nested
True
After:
>>> pl.List(pl.Int8).is_nested()
True
dt.month, dt.weekMost datetime components such as month and week would previously return a UInt32 type. This
has been updated to the smallest appropriate signed integer type. This should reduce memory
consumption.
| Method | Dtype old | Dtype new |
|---|---|---|
| year | i32 | i32 |
| iso_year | i32 | i32 |
| quarter | u32 | i8 |
| month | u32 | i8 |
| week | u32 | i8 |
| day | u32 | i8 |
| weekday | u32 | i8 |
| ordinal_day | u32 | i16 |
| hour | u32 | i8 |
| minute | u32 | i8 |
| second | u32 | i8 |
| millisecond | u32 | i32* |
| microsecond | u32 | i32 |
| nanosecond | u32 | i32 |
*Technically, millisecond can be an i16. This may be updated in the future.
Example
Before:
>>> from datetime import date
>>> s = pl.Series([date(2023, 12, 31), date(2024, 1, 1)])
>>> s.dt.month()
shape: (2,)
Series: '' [u32]
[
12
1
]
After:
>>> s.dt.month()
shape: (2,)
Series: '' [u8]
[
12
1
]
Null data type when no data is presentThis replaces the previous behavior of initializing as a Float32 type.
Example
Before:
>>> pl.Series("a", [None])
shape: (1,)
Series: 'a' [f32]
[
null
]
After:
>>> pl.Series("a", [None])
shape: (1,)
Series: 'a' [null]
[
null
]
replace reimplemented with slightly different behaviorThe new implementation is mostly backwards compatible. Please do note the following:
return_dtype to override the inferred data type, or take advantage of the new function
signature (separate old and new parameters) to influence the return type.Example
Before:
>>> df = pl.DataFrame({"a": [1, 2, 2, 3], "b": [1.5, 2.5, 5.0, 1.0]}, schema={"a": pl.Int8, "b": pl.Float64})
>>> df.select(pl.col("a").replace({2: 100}))
shape: (4, 1)
┌─────┐
│ a │
│ --- │
│ i8 │
╞═════╡
│ 1 │
│ 100 │
│ 100 │
│ 3 │
└─────┘
>>> df.select(pl.struct("a", "b").replace({2: 100}, default=pl.col("b")))
shape: (4, 1)
┌───────┐
│ a │
│ --- │
│ f64 │
╞═══════╡
│ 1.5 │
│ 100.0 │
│ 100.0 │
│ 1.0 │
└───────┘
After:
>>> df.select(pl.col("a").replace({2: 100}))
shape: (4, 1)
┌─────┐
│ a │
│ --- │
│ i64 │
╞═════╡
│ 1 │
│ 100 │
│ 100 │
│ 3 │
└─────┘
>>> df.select(pl.col("a").replace({2: 100}, default=pl.col("b"))) # No struct needed
shape: (4, 1)
┌───────┐
│ a │
│ --- │
│ f64 │
╞═══════╡
│ 1.5 │
│ 100.0 │
│ 100.0 │
│ 1.0 │
└───────┘
value_counts resulting column renamed from counts to countThe resulting struct field for the value_counts method has been renamed from counts to count.
Example
Before:
>>> s = pl.Series("a", ["x", "x", "y"])
>>> s.value_counts()
shape: (2, 2)
┌─────┬────────┐
│ a ┆ counts │
│ --- ┆ --- │
│ str ┆ u32 │
╞═════╪════════╡
│ x ┆ 2 │
│ y ┆ 1 │
└─────┴────────┘
After:
>>> s.value_counts()
shape: (2, 2)
┌─────┬───────┐
│ a ┆ count │
│ --- ┆ --- │
│ str ┆ u32 │
╞═════╪═══════╡
│ x ┆ 2 │
│ y ┆ 1 │
└─────┴───────┘
read_parquet to use Object Store rather than fsspecIf you were using read_parquet, installing fsspec as an optional dependency is no longer
required. The new Object Store implementation was already in use for scan_parquet. It may have
slightly different behavior in certain cases, such as how credentials are detected and how downloads
are performed.
The resulting DataFrame should be identical between versions.
cum* to cum_*Technically, this deprecation was introduced in version 0.19.14, but many users will first
encounter it when upgrading to 0.20. It's a relatively impactful change, which is why we mention
it here.
| Old name | New name |
|---|---|
cumfold |
cum_fold |
cumreduce |
cum_reduce |
cumsum |
cum_sum |
cumprod |
cum_prod |
cummin |
cum_min |
cummax |
cum_max |
cumcount |
cum_count |
empty product returns identity and product ignores nulls
binary, boolean, categorical, date, object, and time selectors (#10806)allow_copy=False (#10822)reversed(df) (#10823)range related functions (#10830)Thank you to all our contributors for making this release possible! @alexander-beedie, @orlp, @reswqa, @ritchie46 and @stinodego
Remove deprecated behavior from vertical aggregations
An upgrade guide is available on our website.
DataFrame init from queries against users' existing database connections (#10649)groupby to group_by (#10656)f64 for rank when method="average" (#10734)all - fix Kleene logic implementation for all/any (#10564)from_arrow to take a generator of RecordBatches, change error type to TypeError (#10529)arange an alias for int_range (#9983)date_range/time_range no longer return a List type (#10526)0.18 (#10527)map to map_batches (#10801)GroupBy.apply to map_groups (#10799)DataFrame.apply to map_rows (#10797)Series/Expr.rolling_apply to rolling_map (#10750)Series/Expr.apply to map_elements (#10678)groupby to group_by (#10656)cut/qcut (#10484)Protocol for interchange classes (#10688)DataFrame init from queries against users' existing database connections (#10649)truncate_ragged_lines (#10660)write_excel arguments (#10589)LazyFrame.collect_async and pl.collect_all_async (#10616)is_in and more generic array construction (#10614)all - fix Kleene logic implementation for all/any (#10564)cast support (#10504)from_arrow to take a generator of RecordBatches, change error type to TypeError (#10529)get_idx_type - use get_index_type instead (#10556)arange an alias for int_range (#9983)date_range/time_range no longer return a List type (#10526)0.18 (#10527)ORDER BY on unselected columns (#10752)pre-wrap instead of pre (#10739)value_counts on column named "counts" (#10737)f64 for rank when method="average" (#10734)on arg type for join_asof (#10690)write_delta (#10633)is_in (#10620)all - fix Kleene logic implementation for all/any (#10564)write_delta with schema in delta_write_options (#10541)pl.Config options relating to shape, column names, and types when rendering HTML (#10449).venv in repo root (#10789)write_database unit tests to properly separate concerns (#10773)adbc release (#10763)connectorx and bump other Python dependencies (#10753)testing docs about module import (#10741)13.0.0 behavior (#10691)sink_parquet docs (#10669)deprecate_renamed_methods util (#10537)inspect.currentframe (#10630)Expr.meta namespace (#10617)Cargo.lock (#10555)make requirements fully refreshes unpinned packages/deps (#10591)expr_dispatch decorator to work on methods with decorators (#10549)pyo3/maturin-action (#10503)Thank you to all our contributors for making this release possible! @JulianCologne, @MarcoGorelli, @Object905, @OndrejSlamecka, @SeanTroyUWO, @VasanthakumarV, @alexander-beedie, @aminalaee, @braaannigan, @c-peters, @ion-elgreco, @lorepozo, @marki259, @mcrumiller, @messense, @orlp, @owrior, @rben01, @reswqa, @ritchie46, @sdamashek, @stinodego, @svaningelgem, @titoeb, @trueb2, @washcycle and @zundertj
This impacts aggregation functions like sum, min, and max. These functions were overloaded to
support both vertical and horizontal computation. Recently, new dedicated functionality for
horizontal computation was released, and horizontal computation was deprecated.
Restore the old behavior by using the horizontal variant, e.g. sum_horizontal.
Example
Before:
>>> df = pl.DataFrame({'a': [1, 2], 'b': [11, 12]})
>>> df.select(pl.sum('a', 'b')) # horizontal computation
shape: (2, 1)
┌─────┐
│ sum │
│ --- │
│ i64 │
╞═════╡
│ 12 │
│ 14 │
└─────┘
After:
>>> df = pl.DataFrame({'a': [1, 2], 'b': [11, 12]})
>>> df.select(pl.sum('a', 'b')) # vertical computation
shape: (1, 2)
┌─────┬─────┐
│ a ┆ b │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞═════╪═════╡
│ 3 ┆ 23 │
└─────┴─────┘
all / anyall will now ignore null values by default, rather than treat them as False.
For both any and all, the drop_nulls parameter has been renamed to ignore_nulls and is now
keyword-only. Also fixed an issue when setting this parameter to False would erroneously result in
None output in some cases.
To restore the old behavior, set ignore_nulls to False and check for None output.
Example
Before:
>>> pl.Series([True, None]).all()
False
After:
>>> pl.Series([True, None]).all()
True
Improving our error messages is an ongoing effort. We did a sweep of our Python code base and made
many improvements to error messages and error types. Most notably, many ValueErrors were changed
to TypeErrors.
If your code relies on handling Polars exceptions, you may have to make some adjustments.
Example
Before:
>>> pl.Series(values=15)
...
ValueError: Series constructor called with unsupported type; got 'int'
After:
>>> pl.Series(values=15)
...
TypeError: Series constructor called with unsupported type 'int' for the `values` parameter
Methods like select and with_columns accept one or more expressions. But they also accept
strings, integers, lists, and other inputs that we try to interpret as expressions. We updated our
internal logic to parse inputs more consistently.
Example
Before:
>>> pl.DataFrame({'a': [1, 2]}).with_columns(None)
shape: (2, 1)
┌─────┐
│ a │
│ --- │
│ i64 │
╞═════╡
│ 1 │
│ 2 │
└─────┘
After:
>>> pl.DataFrame({'a': [1, 2]}).with_columns(None)
shape: (2, 2)
┌─────┬─────────┐
│ a ┆ literal │
│ --- ┆ --- │
│ i64 ┆ null │
╞═════╪═════════╡
│ 1 ┆ null │
│ 2 ┆ null │
└─────┴─────────┘
shuffle / sample now use an internal Polars seedIf you used the built-in Python random.seed function to control the randomness of Polars
expressions, this will no longer work. Instead, use the new set_random_seed function.
Example
Before:
import random
random.seed(1)
After:
import polars as pl
pl.set_random_seed(1)
Creating a consistent and intuitive API is hard; finding the right name for each function, method,
and parameter might be the hardest part. The new version comes with several naming changes, and you
will most likely run into deprecation warnings when upgrading to 0.19.
If you want to upgrade without worrying about deprecation warnings right now, you can add the following snippet to your code:
import warnings
warnings.filterwarnings("ignore", category=DeprecationWarning)
groupby renamed to group_byThis is not a change we make lightly, as it will impact almost all our users. But "group by" is really two different words, and our naming strategy dictates that these should be separated by an underscore.
Most likely, a simple search and replace will be enough to take care of this update:
.groupby(.group_by(apply renamed to map_*apply is probably the most misused part of our API. Many Polars users come from pandas, where
apply has a completely different meaning.
We now consolidate all our functionality for user-defined functions under the name map. This
results in the following renaming:
| Before | After |
|---|---|
Series/Expr.apply |
map_elements |
Series/Expr.rolling_apply |
rolling_map |
DataFrame.apply |
map_rows |
GroupBy.apply |
map_groups |
apply |
map_groups |
map |
map_batches |
Remove deprecated tz\_aware argument
.arr to .list (#8999)DataFrame/LazyFrame (#9008)date_range/ones/zeros to eager=False (#9007).arr to .list (#8999)Utf8 to Decimal. (#9090)repeat (#9046)date_range/ones/zeros to eager=False (#9007)Series declared as int/temporal with floating point values (#9004)time_unit property from Series (#8990)repeat (#9048)arange/date_range/time_range (#9027)DataFrame/LazyFrame (#9008)SQLContext docstring cleanups (#9005).arr to .list (#8999)Thank you to all our contributors for making this release possible! @CloseChoice, @MarcoGorelli, @alexander-beedie, @charliegallop, @jonashaag, @mcrumiller, @raymead, @ritchie46, @sorhawell, @stinodego, @tim-habitat and @universalmind303
don't create duplicate pivot names
toggle_string_cache to enable_string_cache (#7970)sort, top_k, sort_by, and arg_sort_by, raise if descending is a sequence and its length doesn't match the number of columns to sort by (#7957)time_unit/time_zone instead of tu/tz (#7910)struct, concat_str, and arg_sort_by (#7308)shift_and_fill and add default… (#7192)func to function (#7139)Series/Expr methods to keyword-only (#7860)FromParalleIter<Option<str>> for Utf8Chunked ~1.9x (#8058)~2.5x (#8057)~2x. (#8053)into_groups materialization ~-25% (#8036)~25% (#7980)DataFrame init from pyarrow RecordBatch objects, and improve init from Array (#8011)write_ipc to take file=None (returning BytesIO) (#7997)Config methods, reference POLARS_MAX_THREADS in threadpool_size docstring (#7965)struct, concat_str, and arg_sort_by (#7308)sort, top_k, sort_by, and arg_sort_by, raise if descending is a sequence and its length doesn't match the number of columns to sort by (#7957)toggle_string_cache to enable_string_cache (#7970)time_unit/time_zone instead of tu/tz (#7910)shift_and_fill and add default… (#7192)func to function (#7139)Series/Expr methods to keyword-only (#7860)Thank you to all our contributors for making this release possible! @MarcoGorelli, @StefanBRas, @alexander-beedie, @ghuls, @rben01, @ritchie46, @stinodego and @universalmind303
Nothing published for this version
Update Expr.sample signature and change random seeding
Expr.sample signature and change random seeding (#4648)null_equal default to True for Series.series_equal (#5051)Expr.sample signature and change random seeding (#4648)null_equal default to True for Series.series_equal (#5051)Thank you to all our contributors for making this release possible! @Kuhlwein, @braaannigan, @ghuls, @matteosantama, @ritchie46 and @stinodego
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
One column per quarter.