This release is based on the Kotlin 1.9.0.
Removal of Legacy JS target
Some time ago, in Kotlin 1.8, JS IR compiler was promoted to stable and old JS compiler was deprecated.
Kotlin 1.9 promotes the usage of deprecated JS compiler to an error. As a result, kotlinx.serialization no longer builds with the legacy compiler and does not distribute artifacts for it. You can read the migration guide for JS IR compiler here.
Also pay attention to the fact that Kotlin/Native also has some deprecated targets that are going to be removed in the Kotlin 1.9.20. Therefore, kotlinx.serialization 1.6.0-RC and 1.6.0 are likely the last releases that support these targets.
Case insensitivity for enums in Json
This release features a new configuration flag for Json: decodeEnumsCaseInsensitive that allows you to decode enum values in a case-insensitive manner.
For example, when decoding enum class Foo { VALUE_A , VALUE_B} both inputs "value_a" and "value_A" will yield Foo.VALUE_A. You can read more about this feature in the documentation and corresponding PR.
Other bugfixes and enhancements
- Add support to decode numeric literals containing an exponent (#2227) (thanks to Roberto Blázquez)
- Fix NoSuchMethodError related to Java 8 API compatibility (#2328, #2350) (thanks to Björn Kautler)
- Changed actual FormatLanguage annotation for JS and Native to avoid problems with duplicating org.intellij.lang.annotations.Language (#2390, #2379)
- Fix error triggered by 'consume leading class discriminator' polymorphic parsing optimization (#2362)
- Fix runtime error with Serializer for Nothing on the JS target (#2330) (thanks to Shreck Ye)
- Fix beginStructure in JsonTreeDecoder when inner structure descriptor is same as outer (#2346) (thanks to Ugljesa Jovanovic)
- Actualize 'serializer not found' platform-specific message (#2339)
- Fixed regression with serialization using a list parametrized with contextual types (#2331)