This is an update that corresponds to Python Polars 1.41.2.

New features

• <expr>$gather() gains an argument null_on_oob (#1789).

This is an update that corresponds to Python Polars 1.40.1.

New features

• pl$row_index(), a shortcut for pl$int_range(pl$len()) (#1770).
• polars_code_completion_activate() and polars_code_completion_deactivate() to
  enable Polars-specific code completion. This only works in RStudio for now (#1768).
• <expr>$arr$any(), <expr>$arr$all(), <expr>$list$any(), and <expr>$list$all()
  gain an argument ignore_nulls (#1778).
• <dataframe>$merge_sorted() and <lazyframe>$merge_sorted() gain an argument
  maintain_order (#1778).

Other changes

• Bumped rlang dependency to be >= 1.2.0.

This is an update that corresponds to Python Polars 1.39.3.

New features

• <expr>$implode() gains the maintain_order argument to control whether
  the order of elements within each group is preserved (#1751).
• <expr>$set_sorted() gains the nulls_last argument to specify the
  position of null values in the sorted order (#1751).
• <expr>$dt$add_business_days() now accepts a Polars expression for
  the holidays argument, in addition to R Date vectors (#1751).
• $pivot() gains the column_naming argument to control how pivoted
  column names are constructed (#1751).
• New <expr>$truncate() (#1755).
• <expr>$round() can now take mode = "to_zero" (#1755).
• pl$scan_csv() and pl$read_csv() gain the missing_columns argument to
  control behavior when some CSV files have missing columns compared to the
  expected schema (#1754).

Bug fixes

• $unpivot() now properly checks for duplicate column names when using
  custom value_name or variable_name
  (pola-rs/polars#26606, #1751).

This is an update that corresponds to Python Polars 1.38.1.

Deprecations

• The retries argument in scan/read and sink/write functions is deprecated (#1726).
  Use max_retries in storage_options instead.
• The file_cache_ttl argument in pl$scan_csv(), pl$scan_ipc(), pl$scan_ndjson(), and their read_* counterparts is deprecated (#1726).
  Use file_cache_ttl in storage_options instead.
• <expr>$flatten() is deprecated. Use <expr>$list$explode() instead (#1726).

New features

• cs$by_name() gains the expand_patterns argument. When set to TRUE, regex patterns (^...$) and wildcards (*) in column names are expanded (pola-rs/polars#26437, #1726).
• New <expr>$bin$get() to extract a specific byte from a binary value (#1731).
• <expr>$str$split() has two new arguments literal and strict (#1730).
• New pl$scan_lines() and pl$read_lines() to read one or several files into a single column (#1732).
• $sink_ndjson(), $write_ndjson(), $sink_csv(), and $write_csv() can now export compressed files (#1733, #1735).

Bug fixes

• <expr>$rolling_rank_by() now requires the closed argument to be "right" or "both".
  Previously, "left" and "none" were silently accepted but could produce incorrect results (pola-rs/polars#26287, #1726).

Other changes

• The per_partition_sort_by argument of the deprecated partition classes (pl$PartitionByKey(), pl$PartitionMaxSize(), pl$PartitionParted()) has been removed.
  This feature was removed from upstream Polars (pola-rs/polars#26130, #1726).

This is an update that corresponds to Python Polars 1.37.1.

Deprecations

• The experimental partitioning scheme classes (constructed with pl$PartitionByKey(), pl$PartitionMaxSize(), or pl$PartitionParted()) are deprecated in favor of the new experimental partition by class, constructed with pl$PartitionBy() (#1716).

New features

• New S3 method nanoarrow::as_nanoarrow_array_stream(<lazyframe>) to export a LazyFrame via the Arrow C stream interface (pola-rs/polars#25918, #1709).
• <expr>$min_by() and <expr>$max_by() to get the value of a column ordered by another column (#1715).