[KT-48068] Add opt-in support for NSEnum for Kotlin Native iOS via an annotation #5539

stefanhaustein · 2025-10-29T18:22:49Z

Basically implementing the feature suggested here: https://youtrack.jetbrains.com/issue/KT-48068

SvyatoslavScherbina · 2025-10-30T17:09:21Z

/test-private

kotlin-safe-merge · 2025-10-30T17:09:26Z

Failed to process command due to an unexpected exception.

native/objcexport-header-generator/src/org/jetbrains/kotlin/backend/konan/objcexport/stubs.kt

native/native.tests/testData/framework/objcexport/nativeEnum.kt

kotlin-native/runtime/src/main/kotlin/kotlin/native/Annotations.kt

...-generator/impl/k1/src/org/jetbrains/kotlin/backend/konan/objcexport/ObjCExportTranslator.kt

...er/ir/backend.native/src/org/jetbrains/kotlin/backend/konan/objcexport/ObjCExportCodeSpec.kt

...bjcexport-header-generator/src/org/jetbrains/kotlin/backend/konan/objcexport/StubRenderer.kt

SvyatoslavScherbina · 2025-10-31T12:41:03Z

The K2 (Analysis API, aka AA) implementation is missing, which makes the new test fail in this mode.
Run

./gradlew :native:objcexport-header-generator:testAnalysisApi --tests "org.jetbrains.kotlin.backend.konan.tests.ObjCExportHeaderGeneratorTest"

to reproduce.

Related code:

kotlin/native/objcexport-header-generator/impl/analysis-api/src/org/jetbrains/kotlin/objcexport/translateToObjCClass.kt

Line 54 in f90cbf3

this += translateEnumMembers(symbol)

SvyatoslavScherbina · 2025-10-31T12:45:21Z

Another thing to fix: adding a declaration to stdlib requires updating the ABI dump.
Run the following command to do that:

./gradlew :tools:binary-compatibility-validator:cleanTest :tools:binary-compatibility-validator:test --tests "*" -Poverwrite.output=true

stefanhaustein · 2025-11-04T13:27:01Z

The K2 (Analysis API, aka AA) implementation is missing, which makes the new test fail in this mode. Run

I have added this, but AA emits the a comment about the annotation whereas K1 does not... Any suggestions how to resolve this? Current state is the AA output

...end.native/src/org/jetbrains/kotlin/backend/konan/llvm/objcexport/ObjCExportCodeGenerator.kt

...er/ir/backend.native/src/org/jetbrains/kotlin/backend/konan/objcexport/ObjCExportCodeSpec.kt

...-generator/impl/k1/src/org/jetbrains/kotlin/backend/konan/objcexport/ObjCExportTranslator.kt

native/objcexport-header-generator/src/org/jetbrains/kotlin/backend/konan/objcexport/stubs.kt

...-generator/impl/k1/src/org/jetbrains/kotlin/backend/konan/objcexport/ObjCExportTranslator.kt

native/objcexport-header-generator/src/org/jetbrains/kotlin/backend/konan/objcexport/stubs.kt

...-generator/impl/k1/src/org/jetbrains/kotlin/backend/konan/objcexport/ObjCExportTranslator.kt

...er/ir/backend.native/src/org/jetbrains/kotlin/backend/konan/objcexport/ObjCExportCodeSpec.kt

native/native.tests/testData/framework/objcexport/nativeEnum.swift

kotlin-native/runtime/src/main/kotlin/kotlin/native/Annotations.kt

SvyatoslavScherbina · 2025-11-24T14:57:23Z

native/objcexport-header-generator/testData/headers/enumClassWithNamedObjCEnum/Foo.kt

+enum class Foo {
+    // Note that the order is not alphabetic on purpose, ensuring that tests fail if we can't rely on the
+    // order to be preserved for the ordinal value.
+    ALPHA, COPY, BAR_FOO,


There is no test checking how having @ObjCName on enum class or its entries affects the generated NS_ENUM.

I have added a few more cases here -- let me know if I should add any additional specific one...

I tried adding @ObjCName on the literal itself, but that's broken for AA in general, even for "regular" enum literals, and the output won't match, so I can't even document the error easily.

I'll file a bug for this separately

KT-82838 @ObjCName for enum literals inconsistent in AnalysisApi

I tried adding @ObjCName on the literal itself, but that's broken for AA in general, even for "regular" enum literals, and the output won't match, so I can't even document the error easily.

Ah, I see. Thanks!
What can be done: a separate @Test fun can be added to ObjCExportHeaderGeneratorTest for this case and marked with @TodoAnalysisApi. This way, the test will be muted when checking the AA implementation.

I have added annotations to the smaller test and disabled aa there -- I think for the meantime AA is covered by the larger test with enum annotations sufficiently?

...eader-generator/impl/k1/src/org/jetbrains/kotlin/backend/konan/objcexport/ObjCExportNamer.kt

...der-generator/impl/analysis-api/src/org/jetbrains/kotlin/objcexport/translateToObjCHeader.kt

...bjcexport-header-generator/src/org/jetbrains/kotlin/backend/konan/objcexport/StubRenderer.kt

SvyatoslavScherbina

Ok, I think the implementation is good now. But there are a few more things to be done to actually get it merged:

Rebase it to master, squash the commits and write a commit message as recommended here.
Because of KT-82670, an unrelated test has to be adjusted:

--- a/analysis/low-level-api-fir/testData/lazyResolveStdlibSources/wrappedInt.txt
+++ b/analysis/low-level-api-fir/testData/lazyResolveStdlibSources/wrappedInt.txt
@@ -824,7 +824,7 @@ FILE: [ResolvedTo(BODY_RESOLVE)] TypeReference.kt
         }
 
         public open override [ResolvedTo(BODY_RESOLVE)] fun hashCode(): R|kotlin/Int| {
-            ^hashCode this@R|kotlin/jvm/internal/TypeReference|.R|kotlin/jvm/internal/TypeReference.classifier|.R|kotlin/Any.hashCode|().R|kotlin/Int.times|(Int(31)).R|kotlin/Int.plus|(this@R|kotlin/jvm/internal/TypeReference|.R|kotlin/jvm/internal/TypeReference.arguments|.R|kotlin/Any.hashCode|()).R|kotlin/Int.times|(Int(31)).R|kotlin/Int.plus|(this@R|kotlin/jvm/internal/TypeReference|.R|kotlin/jvm/internal/TypeReference.flags|.R|kotlin/Int.hashCode|())
+            ^hashCode this@R|kotlin/jvm/internal/TypeReference|.R|kotlin/jvm/internal/TypeReference.classifier|.R|kotlin/Any.hashCode|().R|kotlin/Int.times|(Int(31)).R|kotlin/Int.plus|(this@R|kotlin/jvm/internal/TypeReference|.R|kotlin/jvm/internal/TypeReference.arguments|.R|kotlin/Any.hashCode|()).R|kotlin/Int.times|(Int(31)).R|kotlin/Int.plus|(this@R|kotlin/jvm/internal/TypeReference|.R|kotlin/jvm/internal/TypeReference.flags|.R|kotlin/Any.hashCode|())
         }
 
         public open override [ResolvedTo(BODY_RESOLVE)] fun toString(): R|kotlin/String| {

I'll run the required tests after that. Hopefully, nothing more will require fixing.
We'll also need some approvals from the codeowners for the changes in stdlib. I'll request them.

stefanhaustein · 2025-12-02T14:40:37Z

Ok, I think the implementation is good now. But there are a few more things to be done to actually get it merged:

Rebase it to master, squash the commits and write a commit message as recommended here.

I think I have rebased the fork successfully but would like to confirm that I won't screw up the branch with the wrong git command.... (I mainly use perforce for work... O:)

Can you confirm that a valid option to squash is to run git rebase -i HEAD~29 in the fork and then replace /every/ "pick" there with "squash" -- including commits from previous times I synced the fork to head? Would you recommend something else?

SvyatoslavScherbina · 2025-12-03T15:36:45Z

I think I have rebased the fork successfully but would like to confirm that I won't screw up the branch with the wrong git command.... (I mainly use perforce for work... O:)

It seems that you actually merged the master branch, not rebased to master.

Can you confirm that a valid option to squash is to run git rebase -i HEAD~29 in the fork and then replace /every/ "pick" there with "squash" -- including commits from previous times I synced the fork to head? Would you recommend something else?

That won't work because of the merge.
First, you need to get rid of the merge commits. Luckily, git rebase helps:

git rebase -i 5952e77c40174e23afb9b2606d3ee3dded3b4b9f

And then you can replace "pick" with "squash" (or "fixup").

...der-generator/test/org/jetbrains/kotlin/backend/konan/tests/ObjCExportHeaderGeneratorTest.kt

native/objcexport-header-generator/testData/headers/enumClassWithObjCEnum/Foo.kt

…Native iOS via an annotation Addresses KT-48068 If not specified otherwise via @ObjCEnum, the NSEnum type name will match the Kotlin Enum ObjC / Swift name with "NSEnum" appended. The new type is accesible via the "nsEnum" property (the same way as the ordinal value via "ordinal")

libraries/stdlib/src/kotlin/annotations/NativeAnnotations.kt

fzhinkin · 2025-12-04T17:54:04Z

libraries/stdlib/src/kotlin/experimental/ExperimentalObjCEnum.kt

+package kotlin.experimental
+
+/**
+ * This annotation marks the experimental [ObjCEnum][kotlin.native.ObjCEnum] annotation.


Please add a description of what this status entails.

Also, is there a path to stabilization?

I have expanded the explanation based on the documentation for ExperimentalNativeApi

Not sure about the stabilization path, I'd expect something along the lines that this would lose experimental status if no issues are observed for a certain amount of time and there is significant usage, but I don't see something similar elsewere...

libraries/stdlib/src/kotlin/annotations/NativeAnnotations.kt

…Annotations.kt Co-authored-by: Filipp Zhinkin <filipp.zhinkin@gmail.com>

stefanhaustein requested review from a team, DmitryNekrasov, fzhinkin, ilya-g and qwwdfsad as code owners October 29, 2025 18:22

stefanhaustein requested review from SvyatoslavScherbina and vsukharev October 29, 2025 18:22

SvyatoslavScherbina reviewed Oct 30, 2025

View reviewed changes

stefanhaustein changed the title ~~Add opt-in support for NSEnum for Kotlin Native iOS via an annotation~~ [KT-48068] Add opt-in support for NSEnum for Kotlin Native iOS via an annotation Nov 4, 2025

stefanhaustein requested a review from a team as a code owner November 4, 2025 13:21

stefanhaustein requested a review from SvyatoslavScherbina November 13, 2025 18:10

SvyatoslavScherbina reviewed Nov 17, 2025

View reviewed changes

stefanhaustein requested a review from SvyatoslavScherbina November 20, 2025 15:06

SvyatoslavScherbina reviewed Nov 24, 2025

View reviewed changes

stefanhaustein requested a review from SvyatoslavScherbina November 26, 2025 18:07

SvyatoslavScherbina previously approved these changes Dec 1, 2025

View reviewed changes

stefanhaustein dismissed SvyatoslavScherbina’s stale review via 9cf092c December 2, 2025 14:06

stefanhaustein requested a review from a team as a code owner December 2, 2025 14:23

stefanhaustein requested a review from dimonchik0036 December 2, 2025 14:23

stefanhaustein requested a review from SvyatoslavScherbina December 3, 2025 13:17

SvyatoslavScherbina reviewed Dec 3, 2025

View reviewed changes

...der-generator/test/org/jetbrains/kotlin/backend/konan/tests/ObjCExportHeaderGeneratorTest.kt Outdated Show resolved Hide resolved

native/objcexport-header-generator/testData/headers/enumClassWithObjCEnum/Foo.kt Outdated Show resolved Hide resolved

stefanhaustein force-pushed the master branch from 4305d7b to 99127a7 Compare December 3, 2025 17:36

stefanhaustein requested a review from SvyatoslavScherbina December 3, 2025 17:38

SvyatoslavScherbina previously approved these changes Dec 4, 2025

View reviewed changes

dimonchik0036 previously approved these changes Dec 4, 2025

View reviewed changes

fzhinkin requested changes Dec 4, 2025

View reviewed changes

Add swiftName documentation

e709855

stefanhaustein dismissed stale reviews from dimonchik0036 and SvyatoslavScherbina via e709855 December 5, 2025 15:41

stefanhaustein and others added 2 commits December 5, 2025 16:43

Improve formatting for libraries/stdlib/src/kotlin/annotations/Native…

16d934e

…Annotations.kt Co-authored-by: Filipp Zhinkin <filipp.zhinkin@gmail.com>

improve the meta-annotation warning

2cc890d

stefanhaustein requested a review from fzhinkin December 5, 2025 15:54

fzhinkin approved these changes Dec 5, 2025

View reviewed changes

[KT-48068] Add opt-in support for NSEnum for Kotlin Native iOS via an annotation #5539

Are you sure you want to change the base?

[KT-48068] Add opt-in support for NSEnum for Kotlin Native iOS via an annotation #5539

Conversation

stefanhaustein commented Oct 29, 2025

Uh oh!

SvyatoslavScherbina commented Oct 30, 2025

Uh oh!

kotlin-safe-merge bot commented Oct 30, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

SvyatoslavScherbina commented Oct 31, 2025

Uh oh!

SvyatoslavScherbina commented Oct 31, 2025

Uh oh!

stefanhaustein commented Nov 4, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

SvyatoslavScherbina Nov 24, 2025

Choose a reason for hiding this comment

Uh oh!

stefanhaustein Nov 26, 2025

Choose a reason for hiding this comment

Uh oh!

stefanhaustein Dec 1, 2025

Choose a reason for hiding this comment

Uh oh!

SvyatoslavScherbina Dec 1, 2025

Choose a reason for hiding this comment

Uh oh!

stefanhaustein Dec 2, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

SvyatoslavScherbina left a comment

Choose a reason for hiding this comment

Uh oh!

stefanhaustein commented Dec 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

SvyatoslavScherbina commented Dec 3, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

fzhinkin Dec 4, 2025

Choose a reason for hiding this comment

Uh oh!

stefanhaustein Dec 5, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

stefanhaustein commented Dec 2, 2025 •

edited

Loading