Python Polars 1.25.2

🏆 Highlights

• Enable common subplan elimination across plans in collect_all (#21747)
• Add lazy sinks (#21733)
• Add PartitionByKey for new streaming sinks (#21689)
• Enable new streaming memory sinks by default (#21589)

🚀 Performance improvements

• Implement linear-time rolling_min/max (#21770)
• Improve InputIndependentSelect by delegating to InMemorySourceNode (#21767)
• Enable common subplan elimination across plans in collect_all (#21747)
• Allow elementwise functions in recursive lowering (#21653)
• Add primitive single-key hashtable to new-streaming join (#21712)
• Remove unnecessary black_boxes in Kahan summation (#21679)
• Box large enum variants (#21657)
• Improve join performance for new-streaming engine (#21620)
• Pre-fill caches (#21646)
• Optimize only a single cache input (#21644)
• Collect parquet statistics in one contiguous buffer (#21632)
• Update Cargo.lock (mainly for zstd 1.5.7) (#21612)
• Don't maintain order when maintain_order=False in new streaming sinks (#21586)
• Pre-sort groups in group-by-dynamic (#21569)

✨ Enhancements

• Add support for rolling_(sum/min/max) for booleans through casting (#21748)
• Support multi-column sort for all nested types and nested search-sorted (#21743)
• Add lazy sinks (#21733)
• Add PartitionByKey for new streaming sinks (#21689)
• Fix replace flags (#21731)
• Add mkdir flag to sinks (#21717)
• Enable joins on list/array dtypes (#21687)
• Add a config option to specify the default engine to attempt to use during lazyframe calls (#20717)
• Support all elementwise functions in IO plugin predicates (#21705)
• Stabilize Enum datatype (#21686)
• Support Polars int128 in from arrow (#21688)
• Use FFI to read dataframe instead of transmute (#21673)
• Enable new streaming memory sinks by default (#21589)
• Cloud support for new-streaming scans and sinks (#21621)
• Add len method to arr (#21618)
• Closeable files on unix (#21588)
• Add new PartitionMaxSize sink (#21573)
• Support engine callback for LazyFrame.profile (#21534)
• Dispatch new-streaming CSV negative slice to separate node (#21579)
• Add NDJSON source to new streaming engine (#21562)
• Support passing token in storage_options for GCP cloud (#21560)

🐞 Bug fixes

• Expose and document partitions (#21765)
• Fix lazy schema for truediv ops involving List/Array dtypes (#21764)
• Fix error due to race condition in file cache (#21753)
• Clear NaNs due to zero-weight division in rolling var/std (#21761)
• Allow init from BigQuery Arrow data containing ExtensionType cols with irrelevant metadata (#21492)
• Disallow cast from boolean to categorical/enum (#21714)
• Don't check sortedness in join_asof when 'by' groups supplied, but issue warning (#21724)
• Incorrect multithread path taken for aggregations (#21727)
• Disallow cast to empty Enum (#21715)
• Fix list.mean and list.median returning Float64 for temporal types (#21144)
• Incorrect (FixedSize)ListArrayBuilder gather implementation (#21716)
• Always fallback in SkipBatchPredicate (#21711)
• New streaming multiscan deadlock (#21694)
• Ensure new-streaming join BuildState is correct even if never fed morsels (#21708)
• IO plugin; support empty iterator (#21704)
• Support nulls in multi-column sort (#21702)
• Window function check length of groups state (#21697)
• Support 128 sum reduction on new streaming (#21691)
• IPC round-trip of list of empty view with non-empty bufferset (#21671)
• Variance can never be negative (#21678)
• Incorrect loop length in new-streaming group by (#21670)
• Right join on multiple columns not coalescing left_on columns (#21669)
• Casting Struct to String panics if n_chunks > 1 (#21656)
• FixFuture attached to different loop error on read_database_uri (#21641)
• Fix deadlock in cache + hconcat (#21640)
• Properly handle phase transitions in row-wise sinks (#21600)
• Enable new streaming memory sinks by default (#21589)
• Always use global registry for object (#21622)
• Check enum categories when reading csv (#21619)
• Unspecialized prefiltering on nullable arrays (#21611)
• Release the gil on explain (#21607)
• Take into account scalar/partitioned columns in DataFrame::split_chunks (#21606)
• Bad null handling in unordered row encoding (#21603)
• Fix deadlock in new streaming CSV / NDJSON sinks (#21598)
• Bad view index in BinaryViewBuilder (#21590)
• Fix CSV count with comment prefix skipped empty lines (#21577)
• New streaming IPC enum scan (#21570)
• Several aspects related to ParquetColumnExpr (#21563)
• Don't hit parquet::pre-filtered in case of pre-slice (#21565)

📖 Documentation

• Add skrub to ecosystem.md (#21760)
• Add example for percentile rank (#21746)
• Make python/rust getting-started consistent and clarify performance risk of infer_schema_length=None (#21734)
• Add expression composability to PySpark comparison (#21473)
• Document read_().lazy() antipattern (#21623)
• Update Polars Cloud interactive workflow examples (#21609)
• Add a Plotnine example to the visualization docs (#21597)
• Add cloud api reference to Ref guide (#21566)

🛠️ Other improvements

• Remove variance numerical stability hack (#21749)
• Only use chrono_tz timezones in hypothesis testing (#21721)
• Remove order check from flaky test (#21730)
• Add sinks into the DSL before optimization (#21713)
• Add missing test case for #21701 (#21709)
• Remove old-streaming from engine argument (#21667)
• Add as_phys_any to PrivateSeries for downcasting (#21696)
• Use FFI to read dataframe instead of transmute (#21673)
• Work around typos ignore bug (#21672)
• Added Test For datetime_range Nanosecond Overflow (#21354)
• Update to edition 2024 (#21662)
• Update rustc (#21647)
• Support object from chunks (#21636)
• Push versioned docs on workflow dispatch (#21630)
• Fail docs early (#21629)
• Check major/minor in docs (#21626)
• Add docs workflow (#21624)
• Add test for 21581 (#21617)
• Remove even more parquet multiscan handling (#21601)
• Remove multiscan handling from new streaming parquet source (#21584)
• Prepare skeleton for partitioning sinks (#21536)

Thank you to all our contributors for making this release possible!
@GaelVaroquaux, @Kevin-Patyk, @MarcoGorelli, @Matt711, @NathanHu725, @alexander-beedie, @coastalwhite, @dependabot[bot], @jrycw, @kdn36, @lukemanley, @mcrumiller, @nameexhaustion, @orlp, @r-brink, @ritchie46, @wence- and dependabot[bot]

Python Polars 1.24.0

🚀 Performance improvements

• Provide a fallback skip batch predicate for constant batches (#21477)
• Parallelize the passing in new streaming multiscan (#21430)

✨ Enhancements

• Add lossy decoding to read_csv for non-utf8 encodings (#21433)
• Add DataFrame.write_iceberg (#15018)
• Add 'nulls_equal' parameter to is_in (#21426)
• Improve numeric stability rolling_{std, var, cov, corr} (#21528)
• IR Serde cross-filter (#21488)
• Give priority to pycapsule interface in from_dataframe (#21377)
• Support writing Time type in json (#21454)
• Activate all optimizations in sinks (#21462)
• Add AssertionError variant to PolarsError in polars-error (#21460)
• Pass filter to inner readers in multiscan new streaming (#21436)

🐞 Bug fixes

• Categorical min/max panicking when string cache is enabled (#21552)
• Don't encode IPC record batch twice (#21525)
• Respect rewriting flag in Node rewriter (#21516)
• Correct skip batch predicate for partial statistics (#21502)
• Make the Parquet Sink properly phase aware (#21499)
• Don't divide by zero in partitioned group-by (#21498)
• Create new linearizer between rowwise new streaming sink phases (#21490)
• Don't drop rows in sinks between new streaming phases (#21489)
• Incorrect lazy schema for Expr.list.diff (#21484)
• Give priority to pycapsule interface in from_dataframe (#21377)
• Duration Series arithmetic operations (#21425)
• Fix unwrap None panic when filtering delta with missing columns (#21453)
• Use stable sort for rolling-groupby (#21444)
• Throw exception if dataframe is too large to be compatible with Excel (#20900)
• Address regression with read_excel not handling URL paths correctly (#21428)

📖 Documentation

• Fix typo (#21554)
• Correct typos and grammar in Python docstrings (#21524)
• Move llm page under misc (#21550)
• Polars Cloud docs (#21548)
• Add LazyFrame.remote docs entry (#21529)
• Specify that the key column must be sorted in ascending order in merge_sorted (#21501)
• Add Polars & LLMs page to the user guide (#21218)
• Mention that statistics=True doesn't enable all statistics in sink_parquet() (#21434)

🛠️ Other improvements

• Don't take ownership of IRplan in new streaming engine (#21551)
• Refactor code for re-use by streaming NDJSON source (#21520)
• Simplify the phase handling of new streaming sinks (#21530)
• Improve IPC sink node parallelism (#21505)
• Use tikv-jemallocator (#21486)
• Rename 'join_nulls' parameter to 'nulls_equal' in join functions (#21507)
• Move rolling to polars-compute (#21503)
• Remove Growable in favor of ArrayBuilder (#21500)
• Introduce a Sink Node trait in the new streaming engine (#21458)
• Add test for rolling stability sort (#21456)
• Add test for empty .is_in predicate filter (#21455)
• Test for unique length on multiple columns (#21418)

Thank you to all our contributors for making this release possible!
@Kevin-Patyk, @MarcoGorelli, @Matt711, @alexander-beedie, @banflam, @braaannigan, @coastalwhite, @dependabot[bot], @etiennebacher, @ghuls, @kevinjqliu, @lukemanley, @mcrumiller, @nameexhaustion, @orlp, @ritchie46, @stijnherfst, @thomasjpfan and dependabot[bot]

Python Polars 1.23.0

🚀 Performance improvements

• Toggle projection pushdown for eager rolling (#21405)
• Fix pathologic rolling + group-by performance and memory explosion (#21403)
• Add sampling to new-streaming equi join to decide between build/probe side (#21197)

✨ Enhancements

• Implement i128 -> str cast (#21411)
• Connect polars-cloud (#21387)
• Version DSL (#21383)
• Make user facing binary formats mostly self describing (#21380)
• Filter hive files using predicates in new streaming (#21372)
• Add negative slicing to new streaming multiscan (#21219)
• Allow iterable of frames as input to align_frames (#21209)
• Implement sorted flags for struct series (#21290)
• Support reading arrow Map type from Delta (#21330)
• Add a dedicated remove method for DataFrame and LazyFrame (#21259)
• Rename credentials parameter to credential in CredentialProviderAzure (#21295)
• Implement merge_sorted for struct (#21205)
• Add positive slice for new streaming MultiScan (#21191)
• Don't take in rewriting visitor (#21212)
• Add SQL support for the DELETE statement (#21190)
• Add row index to new streaming multiscan (#21169)
• Improve DataFrame fmt in explain (#21158)

🐞 Bug fixes

• Method dt.ordinal_day was returning UTC results as opposed to those on the local timestamp (#21410)
• Use Kahan summation for rolling sum kernels. Fix numerical stability issues (#21413)
• Add scalar checks for n and fill_value parameters in shift (#21292)
• Upcast small integer dtypes for rolling sum operations (#21397)
• Don't silently produce null values from invalid input to pl.datetime and pl.date (#21013)
• Allow duration multiplied w/ primitive to propagate in IR schema (#21394)
• Struct arithmetic broadcasting behavior (#21382)
• Prefiltered optional plain primitive kernel (#21381)
• Panic when projecting only row index from IPC file (#21361)
• Properly update groups after gather in aggregation context (#21369)
• Mark test as may_fail_auto_streaming (#21373)
• Properly set fast_unique in EnumBuilder (#21366)
• Rust test race condition (#21368)
• Fix unequal DataFrame column heights from parquet hive scan with filter (#21340)
• Fix ColumnNotFound error selecting len() after semi/anti join (#21355)
• Merge Parquet nested and flat decoders (#21342)
• Incorrect atomic ordering in Connector (#21341)
• Method dt.offset_by was discarding month and year info if day was included in offset for timezone-aware columns (#21291)
• Fix pickling polars.col on Python versions <3.11 (#21333)
• Fix duplicate column names after join if suffix already present (#21315)
• Skip Batches Expression for boolean literals (#21310)
• Fix performance regression for eager join_where (#21308)
• Fix incorrect predicate pushdown for predicates referring to right-join key columns (#21293)
• Panic in to_physical for series of arrays and lists (#21289)
• Resolve deadlock due to leaking in Connector recv drop (#21296)
• Incorrect result for merge_sorted with lexical categorical (#21278)
• Add Int128 path for join_asof (#21282)
• Categorical min/max returning String dtype rather than Categorical (#21232)
• Checking overflow in Sliced function (#21207)
• Adding a struct field using a literal raises InvalidOperationError (#21254)
• Return nulls for is_finite, is_infinite, and is_nan when dtype is pl.Null (#21253)
• Account for minor change in new connectorx release (#21277)
• Properly implement and test Skip Batch Predicate (#21269)
• Infinite recursion when broadcasting into struct zip_outer_validity (#21268)
• Deadlock due to bad logic in new-streaming join sampling (#21265)
• Incorrect result for top_k/bottom_k when input is sorted (#21264)
• UTF-8 validation of nested string slice in Parquet (#21262)
• Raise instead of panicking when casting a Series to a Struct with the wrong number of fields (#21213)
• Defer credential provider resolution to take place at query collection instead of construction (#21225)
• Do not panic in strptime() if format ends with '%' (#21176)
• Raise error instead of panicking for unsupported SQL operations (#20789)
• Projection of only row index in new streaming IPC (#21167)
• Fix projection count query optimization (#21162)

📖 Documentation

• Fix doc for SQL Functions navigation (#21412)
• Fix initial selector example (#21321)
• Add pandas strictness API difference (#21312)
• Improve Expr.name.map docstring example (#21309)
• Add logo to Ask AI (#21261)
• Fix docs for Catalog (#21252)
• AI widget again (#21257)
• Revert plugin (#21250)
• Add kappa ask ai widget (#21243)
• Update social icons in API reference docs (#21214)
• Improve Arrow key feature description (#21171)
• Improve example in IO plugins user guide (#21146)

🛠️ Other improvements

• Move storage of hive partitions to DataFrame (#21364)
• Feature gate merge sorted in new streaming engine (#21338)
• Remove new streaming old multiscan (#21300)
• Add tests for fixed open issues (#21185)
• Try to mimic all steps (#21249)
• Require version for POLARS_VERSION (#21248)
• Fix docs (#21246)
• Avoid unnecessary packaging dependency (#21223)
• Remove unused file (#21240)
• Add use_field_init_shorthand = true to rustfmt (#21237)
• Don't mutate arena by default in Rewriting Visitor (#21234)
• Disable the TraceMalloc allocator (#21231)
• Add feature gate to old streaming deprecation warning (#21179)
• Install seaborn when running remote benchmark (#21168)

Thank you to all our contributors for making this release possible!
@GiovanniGiacometti, @JakubValtar, @MarcoGorelli, @Matt711, @Shoeboxam, @YichiZhang0613, @alexander-beedie, @bschoenmaeckers, @coastalwhite, @edwinvehmaanpera, @erikbrinkman, @etiennebacher, @hemanth94, @henryharbeck, @jqnatividad, @lukemanley, @mcrumiller, @nameexhaustion, @orlp, @r-brink, @ritchie46 and @ydagosto

Python Polars 1.22.0

🚀 Performance improvements

• Reduce sharing in stringview arrays in new-streaming equijoin (#21129)
• Implement native Expr.count() on new-streaming (#21126)
• Speed up list operations that use amortized_iter() (#20964)
• Use Cow as output for rechunk and add rechunk_mut (#21116)
• Reduce arrow slice mmap overhead (#21113)
• Reduce conversion cost in chunked string gather (#21112)
• Enable prefiltered by default for new streaming (#21109)
• Enable parquet column expressions for streaming (#21101)
• Deduplicate buffers again in stringview concat kernel (#21098)
• Add dedicated concatenate kernels (#21080)
• Rechunk only once during join probe gather (#21072)
• Micro-optimise internal DataFrame height and width checks (#21071)
• Speed up from_pandas when converting frame with multi-index columns (#21063)
• Change default memory prefetch to MADV_WILLNEED (#21056)
• Remove cast to boolean after comparison in optimizer (#21022)
• Split last rowgroup among all threads in new-streaming parquet reader (#21027)
• Recombine into larger morsels in new-streaming join (#21008)
• Improve list.min and list.max performance for logical types (#20972)
• Ensure count query select minimal columns (#20923)

✨ Enhancements

• Add projection pushdown to new streaming multiscan (#21139)
• Implement join on struct dtype (#21093)
• Use unique temporary directory path per user and restrict permissions (#21125)
• Enable ingest of objects supporting the PyCapsule interface via from_arrow (#21128)
• Enable new streaming multiscan for CSV (#21124)
• Environment POLARS_MAX_CONCURRENT_SCANS in multiscan for new streaming (#21127)
• Ensure AWS credential provider sources AWS_PROFILE from environment after deserialization (#21121)
• Multi/Hive scans in new streaming engine (#21011)
• Add linear_spaces (#20941)
• IO plugins suppport lazy schema (#21079)
• Add write_table() function to Unity catalog client (#21089)
• Add is_object method to Polars DataType class (#21074)
• Implement merge_sorted for binary (#21045)
• Hold string cache in new streaming engine and fix row-encoding (#21039)
• Add CredentialProviderAzure parameter to accept user-instantiated azure credential classes (#21047)
• Expose unity catalog dataclasses and type aliases (#21046)
• Support max/min method for Time dtype (#19815)
• Implement a streaming merge sorted node (#20960)
• Automatically use temporary credentials API for scanning Unity catalog tables (#21020)
• Add negative slice support to new-streaming engine (#21001)
• Allow for more RG skipping by rewriting expr in planner (#20828)
• Rename catalog schema to namespace (#20993)
• Add functionality to create and delete catalogs, tables and schemas to Unity catalog client (#20956)
• Allow custom JSONEncoder for the json_normalize function, minor speedup (#20966)
• Support passing aws_profile in storage_options (#20965)
• Improved support for KeyboardInterrupts (#20961)
• Make the available concat alignment strategies more generic (#20644)
• Extract timezone info from python datetimes (#20822)
• Add hint for POLARS_AUTO_USE_AZURE_STORAGE_ACCOUNT_KEY to error message (#20942)
• Filter Parquet pages with ParquetColumnExpr (#20714)
• Expose descending and nulls last in window order-by (#20919)

🐞 Bug fixes

• Fix Expr.over applying scale incorrectly for Decimal types (#21140)
• Fix IO plugin predicate with failed serialization (#21136)
• Ensure lit handles datetimes with tzinfo that represents a fixed offset from UTC (#21003)
• Correctly implement take_(opt_)chunked_unchecked for structs (#21134)
• Restore printing backtraces on panics (#21131)
• Use microseconds for Unity catalog datetime unit (#21122)
• Fix incorrect output height for SQL SELECT COUNT(*) FROM (#21108)
• Validate/coerce types for comparisons within join_where predicates (#21049)
• Do not auto-init credential providers if credential fetch returns error (#21090)
• Fix join_where incorrectly dropping transformations on RHS of equality expressions (#21067)
• Quadratic allocations when loading nested Parquet column metadata (#21050)
• Invalidate sortedness flag when sorting from pl.Categorical to pl.Categorical("lexical") (#21044)
• Calling top_k on list type panics (#21043)
• Fix rolling on empty DataFrame panicking (#21042)
• Fix set_tbl_width_chars panicking with negative width (#20906)
• Ensure write_excel recognises the Array dtype and writes it out as a string (#20994)
• Fix merge_sorted producing incorrect results or panicking for some logical types (#21018)
• Fix all-null list aggregations returning Null dtype (#20992)
• Ensure scalar-only with_columns are broadcasted on new-streaming (#20983)
• Improve SQL interface behaviour when INTERVAL is not a fixed duration (#20958)
• Address minor regression for one-column DataFrame passed to is_in expressions (#20948)
• Add Arrow Float16 conversion DataType (#20970)
• Revert length check of patterns in str.extract_many() (#20953)
• Add maintain order for flaky new-streaming test (#20954)
• Allow for respawning of new streaming sinks (#20934)
• Ensure Function name correctness in cse (#20929)
• Don't consume c_stream as iterable (#20899)
• Validate pl.Array shape argument types (#20915)
• Fix from_numpy returning Null dtype for empty 1D numpy array (#20907)
• Consider the original dtypes when selecting columns in write_excel function (#20909)
• Handle boolean comparisons in Iceberg predicate pushdown (#18199)
• Fix map_elements panicking with Decimal type (#20905)

📖 Documentation

• Replace pandas where with mask in Migrating -> Coming from Pandas (#21085)
• Correct Arrow misconception (#21053)
• Add example showing use of write_delta with delta_lake.WriterProperties (#20746)
• Add missing shape param to Array docstring (#20747)
• Add IO plugins to Python API reference (#21028)
• Document IO plugins (#20982)
• Ensure set_sorted description references single-column behavior (#20709)

📦 Build system

• Speed up CI by running a few more tests in parallel (#21057)

🛠️ Other improvements

• Add test for equality filters in Parquet (#21114)
• Add various tests for open issues (#21075)
• Upgrade packages and apply latest formatting (#21086)
• Move python dsl and builder_dsl code to dsl folder (#21077)
• Organize python related logics in polars-plan (#21070)
• Improve binary dispatch (#21061)
• Skip physical order test (#21060)
• Fix new ruff lints (#21040)
• Added test to check for the computation of list.len for null (#20938)
• Add make fix for running cargo clippy --fix (#21024)
• Add tests for resolved issues (#20999)
• Update code coverage workflow to use macos-latest runners (#20995)
• Remove unused arrow file (#20974)
• Deprecate the old streaming engine (#20949)
• Move dt.replace tests to dedicated file, add "typing :: typed" classifier, remove unused testing function (#20945)
• Extract merge sorted IR node (#20939)
• Update copyright year (#20764)
• Move Parquet deserialization to BitmapBuilder (#20896)
• Also publish polars-python (#20933)
• Remove verify_dict_indices_slice from main (#20928)
• Add tests for already resolved issues (#20921)
• Fix the verify_dict_indices codegen (#20920)
• Add ProjectionContext in projection pushdown opt (#20918)

Thank you to all our contributors for making this release possible!
@FBruzzesi, @MarcoGorelli, @aberres, @alexander-beedie, @arnabanimesh, @bschoenmaeckers, @coastalwhite, @deanm0000, @dependabot[bot], @dimfeld, @eitsupi, @etiennebacher, @henryharbeck, @itamarst, @lmmx, @lukemanley, @mcrumiller, @mullimanko, @nameexhaustion, @orlp, @petrosbar, @ritchie46, @siddharth-vi, @skritsotalakis, @taureandyernv and dependabot[bot]

Rust Polars 0.46.0

🏆 Highlights

• Add new Int128Type (#20232)

💥 Breaking changes

• Support writing partitioned parquet to cloud (#20590)

🚀 Performance improvements

• Use BitmapBuilder in yet more places (#20868)
• Make an owned version of append (#20800)
• Use BitmapBuilder in a lot more places (#20776)
• Extend functionality on BitmapBuilder and use in Growables (#20754)
• Specialize first/last agg for simple types in new-streaming engine (#20728)
• Improve state caching and parallelism of window functions (#20689)
• Broadcast without materialization in concat_arr (#20681)
• Cache rolling groups (#20675)
• Use downcast_ref instead of dtype equality in <dyn SeriesTrait as AsRef<ChunkedArray<T>> (#20664)
• Fix performance regression for DataFrame serialization/pickling (#20641)
• Make Parquet verify_dict_indices SIMD (#20623)
• Move to zlib-rs by default and use zstd::with_buffer (#20614)
• Skip filter expansion in eager (#20586)
• Use AtomicWaker in async engine task joiner (#20604)
• Move morsel distribution to the computational async engine (#20600)
• Improve unique pred-pd (#20569)
• Collapse expanded filters in eager (#20493)
• Remove predicate from IR::DataFrame (#20492)
• Add proper distributor to new-streaming parquet reader (#20372)
• Use different binview dedup strategy depending on chunks ratio (#20451)
• Generalize the arg_sort fast path onto Column (#20437)
• Dedup binviews up front (#20449)
• Re-enable common subplan elim for new-streaming engine (#20443)
• Don't collect all LHS arrays in gather (#20441)
• Remove prepare_series for gather kernels (#20439)
• Don't always take all data buffers when gathering views (#20435)
• Order observability optimizations (#20396)
• Purge ChunkedArray Metadata (#20371)
• Drop probe tables in parallel in new-streaming equi-join (#20373)
• Explicit transpose in new-streaming equi-join finalize (#20363)
• Cache dtype on ExprIR (#20331)

✨ Enhancements

• Expose descending and nulls last in window order-by (#20919)
• Add linear_space (#20678)
• Implement df.unique() on new-streaming engine (#20875)
• Add unique operations for Decimal dtype (#20855)
• Add NDJson sink for the new streaming engine (#20805)
• Support nested keys in window functions (#20837)
• Add CSV sink for the new streaming engine (#20804)
• Periodically check python signals ('CTRL-C' handling) (#20826)
• Experimental unity catalog client (#20798)
• Support cumulative aggregations for Decimal dtype (#20802)
• Improve window function caching strategy (#20791)
• Allow different python versions for pickle (#20740)
• Add SQL support for the NORMALIZE string function (#20705)
• Add 'allow_exact_matches' join_asof' (#20723)
• Add new-streaming first/last aggregations (#20716)
• Add Parquet Sink to new streaming engine (#20690)
• Expose IRBuilder (#20710)
• Make automatic use of Azure storage account keys opt-in (#20652)
• Improve GroupsProxy/GroupsPosition to be sliceable and cheaply cloneable (#20673)
• Add str.normalize() (#20483)
• Allow more group_by agg expressions in the new streaming engine (#20663)
• Support writing partitioned parquet to cloud (#20590)
• Add hint to error message for extra struct field in JSON (#20612)
• Add index_of() function to Series and Expr (#19894)
• Update sqlparser-rs, enabling "LEFT" keyword to be optional for anti/semi joins in SQL queries (#20576)
• Add cat.starts_with/cat.ends_with (#20257)
• Add Int128 IO support for csv & ipc (#20535)
• Support arbitrary expressions in 'join_where' (#20525)
• Allow more join lossless casting (#20474)
• Always resolve dynamic types in schema (#20406)
• Order observability optimizations (#20396)
• Add FirstArgLossless supertype (#20394)
• Add dt.replace (#19708)
• Polars build for Pyodide (#20383)
• Add Azure credential provider using DefaultAzureCredential() (#20384)
• Add env var to ignore file cache allocate error (#20356)
• Enable joins between compatible differing numeric key columns (#20332)
• Cache dtype on ExprIR (#20331)
• Serialize DataFrame/Series using IPC in serde (#20266)
• Improve error message on SchemaError (#20326)
• Use better error messages when opening files (#20307)
• Add 'skip_lines' for CSV (#20301)
• Allow subtraction of time dtype columns (#20300)
• Add bin.reinterpret (#20263)
• Allow decoding of non-Polars arrow dictionaries in Arrow and Parquet (#20248)
• Add new Int128Type (#20232)
• IR formatting QoL improvements (#20246)
• Add cat.len_chars and cat.len_bytes (#20211)
• Expose AexprArena (#20230)

🐞 Bug fixes

• Fix from_numpy returning Null dtype for empty 1D numpy array (#20907)
• Fix map_elements panicking with Decimal type (#20905)
• Warn if asof keys not sorted (#20887)
• Avoid name collisions and panicking in object conversion (#20890)
• Incorrect scale used in log and exp for Decimal type (#20888)
• Don't deep clone manuallydrop in GroupsPosition (#20886)
• Fix DuplicateError when selecting columns after join_where or cross join + filter (#20865)
• Incorrect Decimal value for fill_null(strategy="one") (#20844)
• Fix one edge case (out of many) of int128 literals not working (#20830)
• Add height check to frame-level row indexing when key is int (#20778)
• Remove assert that panics on group_by followed by head(n), where n is larger then the frame height (#20819)
• Fix panic InvalidHeaderValue scanning from S3 on Windows (#20820)
• Fix clip for Decimal returning wrong values (#20814)
• Incorrect height from slicing after projecting only the file path column (#20817)
• Shift mask when skipping Bitpacked values in Parquet (#20810)
• Error instead of truncate if length mismatch for several str functions (#20781)
• Support cumulative aggregations for Decimal dtype (#20802)
• Do not print sensitive information to output on POLARS_VERBOSE (#20797)
• Ignore file cache allocation error if fallocate() is not permitted (#20796)
• Incorrect logic in assert_series_equal for infinities (#20763)
• Avoid blocking on async runtime when resolving cloud scans (#20750)
• Fix allow_invalid_certificates being ignored in storage_options (#20744)
• Incorrect output type for map_groups returning all-NULL column (#20743)
• Fix unique(maintain_order=True) raising InvalidOperationError for null array (#20737)
• Don't collapse into a Nested Loop Join if the cross join maintains order (#20729)
• Don't serialize credentials provider (#20741)
• Fix Series.n_unique raising for list of struct (#20724)
• Fix incorrect top-k by sorted column, fix head() returning extra rows (#20722)
• Add outer validity to AnyValueBufferTrusted for structs (#20713)
• Don't partition group-by with non-scalar literals in agg (#20704)
• Incorrect view buffer dedup (#20691)
• Only verify Parquet ConvertedType if no LogicalType is given (#20682)
• Validate length of schema_overrides in read_csv (#20672)
• Fix map_elements ignoring skip_nulls=True for struct dtype (#20668)
• Check for MAP-GROUPS in cloud-eligible (#20662)
• Fix empty output of to_arrow() on filtered unit height DataFrame (#20656)
• Add .default to azure credential provider scope URL (#20651)
• Fix join_asof panicking for invalid tolerance input (#20643)
• Incorrect flag check on is_elementwise (#20646)
• Don't panic but set null type if type is unknown (#20647)
• Fix performance regression for DataFrame serialization/pickling (#20641)
• Fix Int128 dtype serialization (#20629)
• Ensure that SQL LIKE and ILIKE operators support multi-line matches (#20613)
• Properly broadcast in sort_by (#20434)
• Properly load nested Parquet Statistics (#20610)
• AWS environment config was not loaded when credential provider was used (#20611)
• Fix order observability of group-by-dyn (#20615)
• Soundness when loading Parquet string statistics (#20585)
• Fix error filtering after with_columns() on unit height LazyFrame (#20584)
• Restore symbols on Apple by bumping nightly version (#20563)
• Fix variable name in error message for "unsupported data type" in rolling and upsampling operations (#20553)
• Output index type instead of u32 for sum_horizontal with boolean inputs (#20531)
• Fix more global categorical issues (#20547)
• Update eager join doctest on multiple columns (#20542)
• Revert categorical unique code (#20540)
• Add unique fast path for empty categoricals (#20536)
• Fix various Int128 operations (#20515)
• Fix global cat unique (#20524)
• Fix union (#20523)
• Fix rolling aggregations for various integer types (#20512)
• Ensure ignore_nulls is respected in horizontal sum/mean (#20469)
• Fix incorrectly added sorted flag after append for lexically ordered categorical series (#20414)
• More Int128 testing and related fixes (#20494)
• Validate column names in unique() for empty DataFrames (#20411)
• Implement list.min and list.max for list[i128] (#20488)
• Decimal from physical in horizontal min/max and shift (#20487)
• Don't remove sort if first/last strategy is set in unique (#20481)
• Fix join literal behavior (#20477)
• Validate asof join by args in IR resolving phase (#20473)
• Fix align_frames with single row panicking (#20466)
• Allow multiple column sort for Decimal (#20452)
• Fix mode panicking for String dtype (#20458)
• Return correct schema for sum_horizontal with boolean dtype (#20459)
• Properly handle to_physical_repr of nested types (#20413)
• Workaround for mmap crash under Emscripten (#20418)
• Fix using new_columns in scan_csv with compressed file (#20412)
• Fix decimal arithmetic schema (#20398)
• Raise on categorical search_sorted (#20395)
• Don't try to load non-existend List/FSL statistics (#20388)
• Propagate nulls for float methods on all numeric types (#20386)
• Add env var to ignore file cache allocate error (#20356)
• Flip o...

Python Polars 1.21.0

🚀 Performance improvements

• Use BitmapBuilder in yet more places (#20868)
• Make an owned version of append (#20800)
• Use BitmapBuilder in a lot more places (#20776)

✨ Enhancements

• Stabilize methods/functions (#20850)
• Add linear_space (#20678)
• Improve string → temporal parsing in read_excel and read_ods (#20845)
• Implement df.unique() on new-streaming engine (#20875)
• Experimental credential provider support for Delta read/scan/write (#20842)
• Allow column expressions in DataFrame unnest (#20846)
• Auto-initialize Python credential providers in more cases (#20843)
• Add unique operations for Decimal dtype (#20855)
• Add NDJson sink for the new streaming engine (#20805)
• Support nested keys in window functions (#20837)
• Add CSV sink for the new streaming engine (#20804)
• Periodically check python signals ('CTRL-C' handling) (#20826)
• Experimental unity catalog client (#20798)
• Support cumulative aggregations for Decimal dtype (#20802)
• Account for SurrealDB Python API updates (handle both SurrealDB and AsyncSurrealDB classes) in read_database (#20799)
• Drop nest-asyncio in favor of custom logic (#20793)
• Improve window function caching strategy (#20791)
• Support lakefs:// URI for delta scanner (#20757)
• Additional support for loading numpy.float16 values (as Float32) (#20769)

🐞 Bug fixes

• Warn if asof keys not sorted (#20887)
• Ensure explicit values given to column_widths override autofit in write_excel (#20893)
• Avoid name collisions and panicking in object conversion (#20890)
• Incorrect scale used in log and exp for Decimal type (#20888)
• Don't deep clone manuallydrop in GroupsPosition (#20886)
• Fix DuplicateError when selecting columns after join_where or cross join + filter (#20865)
• Incorrect Decimal value for fill_null(strategy="one") (#20844)
• Fix one edge case (out of many) of int128 literals not working (#20830)
• Add height check to frame-level row indexing when key is int (#20778)
• Remove assert that panics on group_by followed by head(n), where n is larger then the frame height (#20819)
• Selectors should raise on + between themselves (#20825)
• Fix panic InvalidHeaderValue scanning from S3 on Windows (#20820)
• Fix clip for Decimal returning wrong values (#20814)
• Incorrect height from slicing after projecting only the file path column (#20817)
• Shift mask when skipping Bitpacked values in Parquet (#20810)
• Error instead of truncate if length mismatch for several str functions (#20781)
• Support cumulative aggregations for Decimal dtype (#20802)
• Allow is_in values to be given as custom Collection (#20801)
• Propagate null instead of panicking in pl.repeat_by() (#20787)
• Do not print sensitive information to output on POLARS_VERBOSE (#20797)
• Ignore file cache allocation error if fallocate() is not permitted (#20796)
• Incorrect logic in assert_series_equal for infinities (#20763)

📖 Documentation

• Update source URL for legislators-historical.csv (#20858)
• Update ML part of ecosystem user guide page (#20596)

🛠️ Other improvements

• Disable 'catalog' in build (#20897)
• Implement negative slice for new streaming IPC (#20866)
• Debloat Series bitops (#20873)
• Reduce python map bloat (#20871)
• Remove todo and test restriction for new-streaming (#20861)
• Dispatch to the in-mem engine for AExpr::Gather (#20862)
• Dispatch to the in-memory engine for multifile sources (#20860)
• Add tests for open issues (#20857)
• Mark 'register_startup' as unsafe (#20841)
• Reduce mode bloat (#20839)
• Rename ContainsMany to ContainsAny (#20785)
• Unpin NumPy in type checking workflow (#20792)
• Add various tests (#20768)
• Small drive-by's (#20772)
• Touch the upload probe for the remote benchmark (#20767)

Thank you to all our contributors for making this release possible!
@alexander-beedie, @arnabanimesh, @braaannigan, @burakemir, @coastalwhite, @etiennebacher, @ion-elgreco, @itamarst, @lukemanley, @mcrumiller, @nameexhaustion, @orlp, @ritchie46 and @stinodego

Python Polars 1.20.0

⚠️ Deprecations

• Make parameter of str.to_decimal keyword-only (#20570)

🚀 Performance improvements

• Extend functionality on BitmapBuilder and use in Growables (#20754)
• Specialize first/last agg for simple types in new-streaming engine (#20728)
• Use PyO3 to convert between Python and Rust datetimes (#20660)
• Improve state caching and parallelism of window functions (#20689)
• Broadcast without materialization in concat_arr (#20681)
• Cache rolling groups (#20675)
• Use downcast_ref instead of dtype equality in <dyn SeriesTrait as AsRef<ChunkedArray<T>> (#20664)
• Fix performance regression for DataFrame serialization/pickling (#20641)
• Make Parquet verify_dict_indices SIMD (#20623)
• Move to zlib-rs by default and use zstd::with_buffer (#20614)
• Skip filter expansion in eager (#20586)
• Improve unique pred-pd (#20569)

✨ Enhancements

• Allow different python versions for pickle (#20740)
• Add SQL support for the NORMALIZE string function (#20705)
• Add 'allow_exact_matches' join_asof' (#20723)
• Add new-streaming first/last aggregations (#20716)
• Add Parquet Sink to new streaming engine (#20690)
• Make automatic use of Azure storage account keys opt-in (#20652)
• Reduce scan_csv() (and friends') memory usage when using BytesIO (#20649)
• Improve GroupsProxy/GroupsPosition to be sliceable and cheaply cloneable (#20673)
• Add str.normalize() (#20483)
• Allow more group_by agg expressions in the new streaming engine (#20663)
• Support loading Excel Table objects by name (#20654)
• Support writing to file objects from write_excel (#20638)
• Raise DuplicateError if given a pyarrow Table object with duplicate column names (#20624)
• Support writing partitioned parquet to cloud (#20590)
• Add hint to error message for extra struct field in JSON (#20612)
• Add index_of() function to Series and Expr (#19894)
• Update sqlparser-rs, enabling "LEFT" keyword to be optional for anti/semi joins in SQL queries (#20576)
• Add cat.starts_with/cat.ends_with (#20257)

🐞 Bug fixes

• Avoid blocking on async runtime when resolving cloud scans (#20750)
• Fix allow_invalid_certificates being ignored in storage_options (#20744)
• Incorrect output type for map_groups returning all-NULL column (#20743)
• Fix unique(maintain_order=True) raising InvalidOperationError for null array (#20737)
• Don't collapse into a Nested Loop Join if the cross join maintains order (#20729)
• Don't serialize credentials provider (#20741)
• Fix Series.n_unique raising for list of struct (#20724)
• Fix incorrect top-k by sorted column, fix head() returning extra rows (#20722)
• Add outer validity to AnyValueBufferTrusted for structs (#20713)
• Don't partition group-by with non-scalar literals in agg (#20704)
• Fix xor operation of selector with Expr (#20702)
• Incorrect view buffer dedup (#20691)
• Only verify Parquet ConvertedType if no LogicalType is given (#20682)
• Validate length of schema_overrides in read_csv (#20672)
• Fix map_elements ignoring skip_nulls=True for struct dtype (#20668)
• Check for MAP-GROUPS in cloud-eligible (#20662)
• Fix empty output of to_arrow() on filtered unit height DataFrame (#20656)
• Add .default to azure credential provider scope URL (#20651)
• Fix join_asof panicking for invalid tolerance input (#20643)
• Incorrect flag check on is_elementwise (#20646)
• Don't panic but set null type if type is unknown (#20647)
• Fix performance regression for DataFrame serialization/pickling (#20641)
• Fix Int128 dtype serialization (#20629)
• Ensure read_excel and read_ods support reading from raw bytes for all engines (#20636)
• Ensure that SQL LIKE and ILIKE operators support multi-line matches (#20613)
• Properly broadcast in sort_by (#20434)
• Properly load nested Parquet Statistics (#20610)
• AWS environment config was not loaded when credential provider was used (#20611)
• Fix order observability of group-by-dyn (#20615)
• Soundness when loading Parquet string statistics (#20585)
• Fix error filtering after with_columns() on unit height LazyFrame (#20584)
• Propagate tenant_id to CredentialProviderAzure if given (#20583)
• Restore symbols on Apple by bumping nightly version (#20563)
• Fix type annotation of str.strip_chars_* methods (#20565)
• Fix variable name in error message for "unsupported data type" in rolling and upsampling operations (#20553)

📖 Documentation

• Add more information for cross joins (#20753)
• Fix typo in sql functions (cosinus -> cosine) (#20676)
• Add links to read_excel "engine_options" and "read_options" docstring (#20661)
• Fix small typo in plugins (polars-dt -> polars-st) (#20657)
• Add polars-h3 and polars-st to plugin list (#20653)
• Add docs reference for Field (#20625)
• Update DataFrame join examples (#20587)
• Miscellaneous minor updates/fixes (#20573)
• Update "group_by_rolling" (deprecated) to "rolling" in user guide (#20548)

📦 Build system

• Update to official release of PyO3 0.23.4 (#20683)
• Officially support Python 3.13 (#20549)

🛠️ Other improvements

• Fix remote benchmark script (#20755)
• Fix tests (#20745)
• Simplify hive predicate handling in NEW_MULTIFILE (#20730)
• Add tests for various open issues (#20720)
• Fixes an Excel test following new fastexcel release (#20703)
• Add tests for various open issues that have been fixed (#20680)
• Don't include debug symbols in benchmark run (#20571)
• Implement CSV, IPC and NDJson in the MultiScanExec node (#20648)
• Don't rely on argument order of optimization_toggle (#20622)
• Fix Python deps installation in remote-benchmark workflow (#20619)
• Fix flaky categorical test (#20591)
• Bump multiversion from 0.7 to 0.8 (#20543)
• Remove unused nested function in LazyFrame.fill_null (#20558)
• Improve bin size info (#20551)

Thank you to all our contributors for making this release possible!
@Jesse-Bakker, @MarcoGorelli, @MoizesCBF, @SamuelAllain, @alexander-beedie, @bschoenmaeckers, @coastalwhite, @eitsupi, @etiennebacher, @itamarst, @jqnatividad, @lukemanley, @mcrumiller, @nameexhaustion, @orlp, @ritchie46 and @stinodego

Python Polars 1.19.0

🚀 Performance improvements

• Collapse expanded filters in eager (#20493)
• Remove predicate from IR::DataFrame (#20492)
• Use different binview dedup strategy depending on chunks ratio (#20451)
• Generalize the arg_sort fast path onto Column (#20437)
• Dedup binviews up front (#20449)
• Re-enable common subplan elim for new-streaming engine (#20443)
• Don't collect all LHS arrays in gather (#20441)
• Remove prepare_series for gather kernels (#20439)
• Don't always take all data buffers when gathering views (#20435)

✨ Enhancements

• Add Int128 IO support for csv & ipc (#20535)
• Support arbitrary expressions in 'join_where' (#20525)
• Allow use of Python types in cs.by_dtype and col (#20491)
• Add an "include_file_paths" parameter to read_excel and read_ods (#20476)
• Allow more join lossless casting (#20474)
• Accept more generic Iterable[bool] in Series.filter (#20431)
• Allow loading data from multiple Excel/ODS workbooks and worksheets (#20465)

🐞 Bug fixes

• Output index type instead of u32 for sum_horizontal with boolean inputs (#20531)
• Fix more global categorical issues (#20547)
• Update eager join doctest on multiple columns (#20542)
• Revert categorical unique code (#20540)
• Add unique fast path for empty categoricals (#20536)
• Fix various Int128 operations (#20515)
• Fix global cat unique (#20524)
• Fix union (#20523)
• Fix rolling aggregations for various integer types (#20512)
• Ensure ignore_nulls is respected in horizontal sum/mean (#20469)
• Fix incorrectly added sorted flag after append for lexically ordered categorical series (#20414)
• More Int128 testing and related fixes (#20494)
• Validate column names in unique() for empty DataFrames (#20411)
• Implement list.min and list.max for list[i128] (#20488)
• Decimal from physical in horizontal min/max and shift (#20487)
• Don't remove sort if first/last strategy is set in unique (#20481)
• Fix join literal behavior (#20477)
• Validate asof join by args in IR resolving phase (#20473)
• Fix align_frames with single row panicking (#20466)
• Allow multiple column sort for Decimal (#20452)
• Fix mode panicking for String dtype (#20458)
• Return correct schema for sum_horizontal with boolean dtype (#20459)
• Fix return type for add_business_days, millennium, century and combine methods in Series.dt namespace (#20436)

📖 Documentation

• Fix typo in DataFrame.cast (#20532)
• Fix flaky doctests (#20516)
• Add examples for bitwise expressions (#20503)
• Clarify the join pre-condition of join_asof (#20509)
• Fix Expr.all description of Kleene logic (#20409)

🛠️ Other improvements

• Increase categorical test coverage (#20514)
• Report wheel sizes (#20541)
• Add tests for floor/ceil on integers (#20479)
• Expose and rewrite 'can_pre_agg' (#20450)
• Skip test on windows; kuzu import segfaults (#20463)
• Add a TypeCheckRule to the optimizer (#20425)

Thank you to all our contributors for making this release possible!
@Biswas-N, @IndexSeek, @Prathamesh-Ghatole, @Terrigible, @alexander-beedie, @brifitz, @coastalwhite, @dependabot, @dependabot[bot], @jqnatividad, @lukemanley, @mcrumiller, @orlp, @ritchie46 and @siddharth-vi

Python Polars 1.18.0

🏆 Highlights

• Add new Int128Type (#20232)

🚀 Performance improvements

• Order observability optimizations (#20396)
• Purge ChunkedArray Metadata (#20371)
• Explicit transpose in new-streaming equi-join finalize (#20363)
• Cache dtype on ExprIR (#20331)
• Lower overhead for BytecodeParser on introspection of incompatible UDFs (#20280)

✨ Enhancements

• Always resolve dynamic types in schema (#20406)
• Support loading data from multiple Excel/ODS workbooks (#20404)
• Add "drop_empty_cols" parameter for read_excel and read_ods (#20430)
• Order observability optimizations (#20396)
• Add FirstArgLossless supertype (#20394)
• Add dt.replace (#19708)
• Polars build for Pyodide (#20383)
• Add Azure credential provider using DefaultAzureCredential() (#20384)
• Add env var to ignore file cache allocate error (#20356)
• Enable joins between compatible differing numeric key columns (#20332)
• Cache dtype on ExprIR (#20331)
• Serialize DataFrame/Series using IPC in serde (#20266)
• Improve error message on SchemaError (#20326)
• Use better error messages when opening files (#20307)
• Add 'skip_lines' for CSV (#20301)
• Allow subtraction of time dtype columns (#20300)
• Add bin.reinterpret (#20263)
• Allow decoding of non-Polars arrow dictionaries in Arrow and Parquet (#20248)
• Streamline creation of empty frame from Schema (#20267)
• Add cat.len_chars and cat.len_bytes (#20211)
• Expose AexprArena (#20230)

🐞 Bug fixes

• Fix nullable object in map_elements (#20422)
• Properly handle to_physical_repr of nested types (#20413)
• Properly raise UDF errors (#20417)
• Workaround for mmap crash under Emscripten (#20418)
• Fix using new_columns in scan_csv with compressed file (#20412)
• Fix return type of Series.dt.add_business_days (#20402)
• Fix decimal series dispatch (#20400)
• Fix decimal arithmetic schema (#20398)
• Raise on categorical search_sorted (#20395)
• Fix plotting f-strings and docstrings (#20399)
• Don't try to load non-existend List/FSL statistics (#20388)
• Propagate nulls for float methods on all numeric types (#20386)
• Add env var to ignore file cache allocate error (#20356)
• Flip order on right join (#20358)
• Correctly parse special float values in from_repr (#20351)
• Fix incorrect object store caching for ADLS URI (#20357)
• Use the same encoding for nullable as non-nullable arrays (#20323)
• Improve error message on SchemaError (#20326)
• Boolean optional slice pushdown (#20315)
• Properly handle from_physical for List/Array (#20311)
• Ignore quotes in csv comments (#20306)
• Ensure pl.datetime returns empty column when input columns are empty (#20278)
• Ensure output height does not change on lazy projection pushdown with aggregations (#20223)
• Fix error writing on Windows to locations outside of C drive (#20245)
• Incorrect comparison in some cases with filtered list/array columns (#20243)
• Ensure height is maintained in SQL SELECT 1 FROM (#20241)
• Properly account for updated Categorical in .unique() kernel (#20235)

📖 Documentation

• Improve docstring clarity (#20416)
• Update GPU engine installation instructions to remove --extra-index-url from CUDA 12 packages (#20381)
• Remove Plugins overview page without information (#20348)
• Small fixes/clarifications in user guide (#20335)
• Improve docs about NaN (#20310)
• Fix substr function param definition (#19054)
• Include parquet options in BigQuery I/O write sample (#20292)
• Fix typo in fork warning (#20258)

📦 Build system

• Add project.dynamic = ["version"] to pyproject.toml (#20345)
• Update pyo3 and numpy crates to version 0.23 (#20111)
• Build wheels for ARM Windows in Python release workflow (#20247)

🛠️ Other improvements

• Enable masked out list, struct and array elements in parametric tests (#20365)
• Move hive partitioning/multi-file handling outside of readers (#20203)
• Purge ChunkedArray Metadata (#20371)
• Correcting misspelled return value and unifying regional spelling (#20375)
• Add test for select(len()) (#20343)
• Make parametric tests include pl.List and pl.Array by default (#20319)
• Use Column in Row Encoding (#20312)
• Don't warn on fork hook (#20309)
• Don't deconstruct CsvParseOptions (#20302)
• Allow decoding of non-Polars arrow dictionaries in Arrow and Parquet (#20248)
• Prepare test suite for Python 3.13 support (#20297)
• Add FunctionCastOptions and conservative IR-level cast type-checking (#20286)
• Add more descriptive error message for failure of vstack/extend (#20299)
• Clean up some remnants of Python 3.8 support (#20293)
• Add new Int128Type (#20232)
• Add test for BytesIO overwritten after scan (#20240)
• Expose AexprArena (#20230)

Thank you to all our contributors for making this release possible!
@Jesse-Bakker, @Terrigible, @ZemanOndrej, @alexander-beedie, @balbok0, @beckernick, @bschoenmaeckers, @coastalwhite, @georgestagg, @hamdanal, @haocheng6, @kszlim, @lukemanley, @mcrumiller, @nameexhaustion, @noexecstack, @orlp, @ptiza, @r-brink, @ritchie46, @rodrigogiraoserrao, @stijnherfst, @stinodego, @tswast and @zero-stroke

Python Polars 1.17.1

🐞 Bug fixes

• Fix incorrect lazy select(len()) with some select orderings (#20222)
• Fix assertion panic on LazyFrame scratch.is_empty() (#20219)

Thank you to all our contributors for making this release possible!
@nameexhaustion and @ritchie46