<@UBTGMFRCZ> <@U014HUWN7B5> <https://github.com/google/ksp/issues/1463> Was this properly fixed for ...

# ksp

[Paul Woitaschek]

@cb @Ting-Yuan Huang
https://github.com/google/ksp/issues/1463
Was this properly fixed for you in 1.0.13? I can’t find a linked commit and we’re facing the same issue with 1.0.13

[cb]

Seems to be! I re-enable configuration cache and everything is working

[Paul Woitaschek]

For us the ios ksp tasks started becoming skipped because of their onlyIf condition:

Maybe our setup is a little special as the sources live in appleMain which are processed

[Paul Woitaschek]

Hm maybe this one has a caching problem because we conditionally enable / disable targets:

https://github.com/google/ksp/blob/e43fde17c4cd6fe87d667b5d3af78f40c4cb1926/gradle[…]c/main/kotlin/com/google/devtools/ksp/gradle/KotlinFactories.kt

[Paul Woitaschek]

taskName: kspKotlinWatchosArm64
kspTask.konanTarget.enabledOnCurrentHost: false
HosManager#enabled=[linux_x64, linux_arm32_hfp, linux_arm64, linux_mips32, linux_mipsel32, mingw_x86, mingw_x64, android_x86, android_x64, android_arm32, android_arm64, wasm32, macos_x64, macos_arm64, ios_arm32, ios_arm64, ios_x64, ios_simulator_arm64, watchos_arm32, watchos_arm64, watchos_x86, watchos_x64, watchos_simulator_arm64, watchos_device_arm64, tvos_arm64, tvos_x64, tvos_simulator_arm64]
konanTarget=watchos_arm64
HostManager#isEnabled=false

What?

[Paul Woitaschek]

Something is really really weird here:

This prints:

match=ios_arm64. testTarget=ios_arm64. I’m sane=false

[Paul Woitaschek]

Theory:
The KonanTarget class is marked as Serializable. However the entries are no 1.9 data objects but regular objects.
This leads to the equals function only using object identity. After deserialization from config cache, that always evaluates to false.

[Paul Woitaschek]

@Ting-Yuan Huang
I’m not sure exactly how I could file a good bug report here. Do you have enough information to file an issue?

[Ting-Yuan Huang]

@Paul Woitaschek Would you mind giving me a little test case? I'd like to do some more investigation and have a better understanding before filing it to JetBrains.

[Paul Woitaschek]

@Ting-Yuan Huang
@yigit
https://github.com/google/ksp/issues/1522

I think that’s the same thing here, right?

[Paul Woitaschek]

And up:
https://github.com/google/ksp/pull/1523