Hey, while we waiting for multiplatform K2, I was ...
# k2-early-adopters
o
Hey, while we waiting for multiplatform K2, I was able to hack
rsocket-kotlin
to be jvm project instead of multiplatform: https://github.com/rsocket/rsocket-kotlin/tree/k2 And configured CI to run with and without K2 with different kotlin versions (CI run example here: https://github.com/rsocket/rsocket-kotlin/actions/runs/2914307633) And found 2 issues already: • https://youtrack.jetbrains.com/issue/KT-53698/K2-fails-to-resolve-suppressed-internal-symbolhttps://youtrack.jetbrains.com/issue/KT-53699/K2-Exception-during-IR-lowering-in-code-with-coroutines Will continue creating issues if something new will happen Also will try to setup similar setup for JS and Native, when K2 for them will be ready, until K2 multiplatform support P.S. if K2 for JS/Native is already available with some highly experimental flag, please point on it, so I will setup such CI runs for them 🙂
👍 3
👌 3
d
Thank you for your feedback JS and Native are not ready yet even in dev builds, but there is a possibility that JS support will be included in 1.8 (it's more or less ready, but only in branch)
o
ok! will wait for it will be ready to test it, right after dev build released 🙂
minor question: as I will not be able to test K2 locally (because of mpp nature of project) and run tests periodically on CI, which data will be more accurate/useful: 1. with or without incremental compilation (mostly it will be clean build, so incremental compilation will mainly increase overall time in my tests) 2. if incremental - use classpath snapshoting or not? 3. enable or disable gradle cache - will it affect anything?