I'm playing around with Kotlin 1.3 and IntellJ is ...

# intellij

[bartvh]

I'm playing around with Kotlin 1.3 and IntellJ is giving me lots of these warnings now: Class 'kotlin.collections.CollectionsKt__CollectionsJVMKt' is compiled by a pre-release version of Kotlin and cannot be loaded by this version of the compiler. Using Gradle, the code compiles and runs fine. Could it be because I'm using plugin version 1.3-rc146 but in Gradle using 1.3-M2 (which I must because of a bug in the newer version)?

[dsavvinov]

in Gradle using 1.3-M2

Yes, that's exactly the reason. Can you please point to the issue in tracker (maybe we can suggest some build in-between 1.3-M2 and 1.3-rc-146 without the issue)?

[bartvh]

Yes I've been planning to do that. However, considering that there may always be valid reasons to use a Kotlin version other than the latest, is there some way I can rollback IntelliJ to the 1.3-M2 plugin?

[bartvh]

Btw, tl;dr; the bug is that someKClass.primaryConstructor.callBy(someMap) now always throws an internal exception

[bartvh]

However, I haven't reported this yet because I wanted to reproduce it in a standalone code sample that is not private

[dsavvinov]

Milestone releases are not supposed to be used for a long time, that's why they produce binaries which are marked as "pre-release". Release compilers (1.3.X, as well as RCs) will always reject such binaries, so if there are some people who depend on your binaries, I highly suggest you to move to some stable version.

Also, if something works in 1.3-M2, but doesn't works in 1.3-rc-XXX, it's quite bad regression. If you're not able to post minimized reproduction right now, then could you please post an exception stacktrace at least? We'll look into it

[bartvh]

I'm working on a repro right now, and will post today regardless of whether I can get it done or not 👍

[bartvh]

Alright I've found the issue

[bartvh]

https://youtrack.jetbrains.net/issue/KT-27598

[dsavvinov]

Thanks a lot for the minimized reproduction!
1.3-RC introduced mangling of functions which accept inline classes; looks like we've missed this case in the reflection support.

[bartvh]

No problem! By the way, mangling was introduced in M2 already, just saying 😛 Anyway, doesn't matter as long as it works in 1.3.0 👍

[dsavvinov]

Indeed 😂