icu_datetime

The ICU4X Technical Committee is happy to announce ICU4X 2.2, an update to our modular, portable, and secure i18n library.

Important changes in ICU4X 2.2 include:

1. Latest i18n data: This release includes an update to CLDR 48.2 and support for TZDB 2026a.
2. New and improved icu_calendar: This release contains new APIs in icu_calendar, as well as some behavior changes in icu_calendar; see the migration notes section.
   1. Stable datetime arithmetic: It is now possible to to add and subtract dates via date.try_add_with_options() and date.try_until_with_options().
   2. More flexible date construction: The new Date::try_from_fields() API allows constructing dates from all kinds of constituent data: extended years, era years, ordinal months, month codes, etc., and supports different kinds of overflow handling.
   3. Typed months: The new Month type replaces month codes in a type safe way. Date::try_from_codes() has been replaced with Date::try_new() which uses these better types. Various calendar-specific constructors for lunisolar calendars have also been refreshed.
   4. Experimental third-party crate integration: We now support converting from and formatting types from the jiff, chrono, and time crates. See icu_datetime::input::third_party. We’re not yet sure if these integrations should live in ICU4X, in the third party crates, or some adapter crate, and are open to feedback!
3. Experimental Kotlin Bindings: We now have Kotlin bindings for ICU4X (found under ffi/mvn), with the same set of supported APIs as our other cross-language bindings.
4. Experimental features:
   1. Display names: Adds new internal data layout exposed via RegionDisplayName and ScriptDisplayName APIs. The old data layout, optimized for loading multiple names at once, is moved into the multi module. Please share feedback on our tracking issue.
   2. Compact decimal formatter: Please share feedback in preparation for stabilization in a future release.
   3. ML segmentation: Initial code for RAdaBoost word segmenter for Chinese and CNN word segmenter for Thai.
5. Unstable Cargo feature: Replaced the "experimental" Cargo feature with an "unstable" Cargo feature.
6. Better hour cycles: Adds support for Clock12 and Clock24 in datetime formatting.

Migration Notes

1. The JapaneseExtended calendar has been deprecated. CLDR removed the pre-Meiji eras because no implementation correctly implemented them. Clients of pre-Meiji eras are encouraged to comment on CLDR-11400 and potentially write a custom east_asian_traditional::Rules implementation of the pre-Meiji lunisolar Japanese calendar.

2. Hijri<AstronomicalSimulation> has also been deprecated. It previously returned estimates of the Hijri calendar based on simulations published by Reingold et al. To restore this behavior, write a custom hijri::Rules implementation using calendrical_calculations.

See the full changelog for more information

1.3.2 was released to clean up the range dependency on zerovec deliberately introduced in 1.3.0. It includes all previous 1.3.x changes

A subset of crates received a 1.3.1 patch release, to incorporate documentation fixes (https://github.com/unicode-org/icu4x/pull/4096, https://github.com/unicode-org/icu4x/pull/4097, https://github.com/unicode-org/icu4x/pull/4099, https://github.com/unicode-org/icu4x/pull/4101). These crates were: icu_calendar, icu_casemap, icu_datetime, icu_locid_transform, icu_provider.

• General data model
  • Non-exhaustive errors for locid, calendar, decimal, plurals (unicode-org#1792, unicode-org#1793)
  • Rename "serialize" feature to "serde" (unicode-org#1797)
  • Turn all errors into Copy types (unicode-org#1657)
  • Components
    • calendar:
      • Coptic, Indian and Ethiopian calendars (unicode-org#1660, unicode-org#1715, unicode-org#1779)
      • Calendar arithmetic (unicode-org#1614)
    • datetime:
      • Formatting for fractional seconds (unicode-org#1813, unicode-org#1801)
      • Support for day of week in month ('F') (unicode-org#1770)
      • Custom fallbacking for TimeZoneFormatter (unicode-org#1591)
      • Support for week-of-month (unicode-org#1468)
      • Bug fix to get_best_available_format_pattern skeleton matching logic (unicode-org#1549)
    • decimal: No updates
    • locale_canonicalizer:
      • ZeroCopy support (unicode-org#1760, unicode-org#1777)
    • locid:
      • Simplified language representation (unicode-org#1695)
      • Region, Script and Variant subtags ULE (unicode-org#1696)
    • plurals:
      • Update data model to use ZeroVec (unicode-org#1240)
    • properties:
      • Bidi support (unicode-org#1716, unicode-org#1784)
  • Utilities
    • codepointtrie:
      • Use 0 for error value for Rust out-of-bounds for primitive trie value types (unicode-org#1804)
    • crlify: New util for line ending conversions
    • deduplicating_array: No updates
    • fixed_decimal:
      • Improvements to FixedDecimal f64 APIs (unicode-org#1718)
    • litemap:
      • Pluggable LiteMap backends (unicode-org#1769)
    • pattern: No updates
    • uniset: No updates
    • writeable: No updates
    • yoke: No updates
    • zerofrom: No updates
    • zerovec:
      • ZeroVec derive improvements (unicode-org#1780)
      • Support non-Ord values in ZeroMap (unicode-org#1743)
      • Add OptionULE and OptionVarULE (unicode-org#1736)
      • Rename ZeroVec::from_slice and add new method for const-constructed ZeroSlice (unicode-org#1728)
      • Treat ZeroMap sort order as an optional invariant (unicode-org#1727)
      • Add ZeroMap::get_copied_by (unicode-org#1722)
      • Generalize PairULE to support longer tuples (unicode-org#1721)
      • Add more AsULE impls for primitives (unicode-org#1672)
      • Add cast methods to ZeroVec and ZeroSlice (unicode-org#1651)
      • Add RawBytesULE::slice_from_byte_slice (unicode-org#1648)
      • Create façades for ZeroVec types, hide internal code organization modules (unicode-org#1629)
      • Add zerovec::skip_kv and zerovec::skip_ord attributes, as well as generalized attribute handling framework (unicode-org#1613)
      • Rename as_unaligned to to_unaligned (unicode-org#1619)
  • FFI:
    • Updating to Diplomat 0.3
    • Making testdata an optional FFI dep (unicode-org#1820)
    • Split out capi targets: make separate freertos, staticlib, and cdylib crates as targets (unicode-org#1747)
  • Experimental:
    • crabbake: Initial version of baked data provider (unicode-org#1825)
    • segmenter:
      • Support production-ready data provider for segmenters (unicode-org#1652)
      • Implement dictionary based segmenter for line segmenter. (unicode-org#1644)
      • Wire DataProvider into UAX29 segmenters (unicode-org#1627)
      • Move UAX#14 defines to line.toml (unicode-org#1568)
      • Add segmenter factories to generate UAX29 iterators (unicode-org#1602)

• General data model
  • DataPayload no longer needs a lifetime (unicode-org#1297, unicode-org#1279)
  • Re-write DataKey (unicode-org#1511)
  • Rewrite ErasedDataProvider as AnyProvider (unicode-org#1495)
  • Add EitherProvider and rename IterableDataProviderCore to IterableProvider (unicode-org#1455)
  • Change DataRequest to be borrowed in BufferProvider (unicode-org#1416)
  • Replace SerdeDeDataProvider with BufferProvider (unicode-org#1369, unicode-org#1384)
  • Components
    • calendar:
      • Julian, Japanese, and Buddhist calendars (unicode-org#1351, unicode-org#1394, unicode-org#1305)
      • DateTimeFormat integration (unicode-org#1339)
      • Bugfix around arithmetic (unicode-org#1352)
    • datetime:
      • Week-of-year support (unicode-org#1206)
      • DateTimeFormat::resolve_components() (unicode-org#1362)
      • Era formatting (unicode-org#1346)
      • TimeZoneFormatterConfig (unicode-org#1256)
      • New data model for organizing calendar data (unicode-org#1300)
      • Bugfix around missing localized strings in time zone data (unicode-org#1405)
    • decimal: No updates
    • locale_canonicalizer:
      • Bugfix in maximization (unicode-org#1171)
      • Update data model to use LiteMap (unicode-org#1275)
    • locid: No updates
    • plurals:
      • Update data model to use ZeroVec (unicode-org#1240)
    • properties:
      • Rename resource key category for properties (unicode-org#1406)
      • Rename enums for General_Category (unicode-org#1355)
      • Implement the Canonical_Combining_Class property (unicode-org#1347)
      • Implement Script_Extensions property (unicode-org#1353)
      • Add General_Category predicate functions (unicode-org#1310)
      • Implement Grapheme_Cluster_Break, Word_Break, and Sentence_Break Unicode properties (unicode-org#1233)
  • Utilities
    • codepointtrie: No changes
    • deduplicating_array: New utility for efficient serialized representation of data with duplicates
    • fixed_decimal:
      • Padding and truncation APIs (unicode-org#1482, unicode-org#1507, unicode-org#1195)
      • Add double-to-decimal via ryū (unicode-org#1217)
      • Handle exponents in FixedDecimal::from_str() (unicode-org#1265)
    • litemap:
      • Add LiteMap::get_indexed() and LiteMap::find_index()
      • Handle serialization of tuples (etc) in litemaps (unicode-org#1306)
    • pattern: No updates
    • uniset: No updates
    • writeable:
      • Adding parts functionality to Writeable (unicode-org#1438)
      • Change Writeable::writeable_to_string to return a Cow (unicode-org#1452)
      • Implementing Writeable for all integers (unicode-org#1408)
      • Making writeable::LengthHint a range (unicode-org#1400)
      • Simplifying assert_writeable_parts_eq and assert_writeable_eq (unicode-org#1522, unicode-org#1399)
    • yoke:
      • Remove Yokeable::Output from ZeroCopyFrom trait (unicode-org#1499)
      • Add EitherCart (unicode-org#1484)
      • Remove attach_to_option_cart() (unicode-org#1348)
      • Homogenize yoke generic impls to always work with Yokeable, add OwnedYokeable (unicode-org#1302)
    • zerovec:
      • Move over to a model where the vector types deref to ZeroSlice and VarZeroSlice (unicode-org#1418, unicode-org#1371)
      • Simplify ZeroVec/VarZeroVec error handling, consolidate ULEError type (unicode-org#1389)
      • Simplify ZeroMapKV (unicode-org#1334)
      • Add ZeroMap2d, a two-dimensional zero-copy map (unicode-org#1432)
      • Add borrowed-only version of ZeroMap (unicode-org#1238)
      • Add various helper functions to zerovec (unicode-org#1430)
      • Rename PlainOldULE to RawBytesULE (unicode-org#1413)
      • Improve EncodeAsVarULE (unicode-org#1385)
      • Add EncodeAsVarULE for ZeroVec (unicode-org#1274, unicode-org#1407)
      • Various trait impls (unicode-org#1332, unicode-org#1330, unicode-org#1328, unicode-org#1287)
      • Document ULE alignment guarantee; update all impls with checklists (unicode-org#1294)
      • Fix PairULE validation function (unicode-org#1266)
  • FFI:
    • Updating to the latest Diplomat
    • Add FFI for constructing Data Structs, including decimal data structs (unicode-org#1497)
    • Add padding/truncation to FFI (unicode-org#1501)
    • Add FFI for constructing fixed decimals from float (unicode-org#1483)
    • Properties FFI (unicode-org#1269)
  • Experimental:
    • New ListFormatter experiment
    • More progress on segmentation experiment

• Updated to CLDR 40 (unicode-org#1216)
  • Functional Unicode property APIs and data; moved to new crate icu_properties (unicode-org#148)
  • DateTimeFormat optimization: memory usage cut by two-thirds and perf improved by 40% (unicode-org#519)
  • BlobDataProvider allowing for dynamic, asynchronous data loading (unicode-org#1084)
  • Major upgrades to the zerovec utility crate and new benchmarks for ZeroMap (unicode-org#1082, unicode-org#1087)
  • Initial support for time zones in DateTimeFormat components bag (unicode-org#845)
  • Bug fixes in LocaleCanicalizer (unicode-org#1189, etc.)
  • Week of month/year arithmetic (unicode-org#918)
  • Major improvements on code size across FFI (unicode-org#962)
  • Iterations on ICU4X Segmenter (unicode-org#1045, unicode-org#1165, etc.)
  • Experimental ListFormatter (unicode-org#1053)
  • ICU4X now requires Rust 1.56 (unicode-org#1201)

• Static data provider without filesystem dependency (unicode-org#78)
  • Infrastructure around zero-copy deserialization, including ZeroVec and Yoke (unicode-org#652, unicode-org#667)
  • Most ICU4X components support no_std (unicode-org#812)
  • Hour cycle preferences are used by DateTimeFormat (unicode-org#671)
  • Skeleton matching supports additional field widths (unicode-org#584)
  • Add canonicalize method to LocaleCanonicalizer (unicode-org#747)
  • Added range iterator on UnicodeSet (unicode-org#836)
  • Add PluralRules::categories() function (unicode-org#789)
  • Consolidated data generation tool into new binary icu4x-datagen

• Updated CLDR to 39 (unicode-org#640)
  • Initial release of:
    • icu_decimal (unicode-org#590)
    • icu_locale_canonicalizer (unicode-org#417)
  • DateTimeFormat gets support for
    • day periods (unicode-org#435)
    • time zones (unicode-org#418)
    • components bag (unicode-org#481)
  • UnicodeSet gets preliminary support for L3a subset (unicode-org#478)
  • PluralRules support E operand (unicode-org#407)
  • New utilities:
    • Writeable
    • ZeroVec
    • LiteMap
    • FixedDecimal
    • Pattern
  • Early prototype of C API - icu_capi

• Initial release of ICU4X.
  • Initial release of:
    • icu
    • icu_locid
    • icu_plurals
    • icu_datetime
    • icu_uniset
    • icu_provider
    • icu_provider_fs
    • icu_provider_cldr
    • icu_testdata
    • icu4x_ecma402
    • fixed_decimal