Now you list all the needed artifacts right in the integration. It is convenient, but not efficient enough. It's better to publish artifacts for each spark/scala variation and use only a single dependency in the integration. It will significantly speed up the resolution process (3-6 times). I will improve the resolution of multiple dependencies on our side too.
Hmm, that would result in another 20 packages on maven central called org.jetbrains.kotlinx.spark.jupyter-dependencies_x.x.x_2.x. I don't think Sonatype can handle that XD The current 40+ are already an issue
Now you list all the needed artifacts right in the integration. It is convenient, but not efficient enough. It's better to publish artifacts for each spark/scala variation and use only a single dependency in the integration. It will significantly speed up the resolution process (3-6 times). I will improve the resolution of multiple dependencies on our side too.