Dávid Szakállas

This would speed up GitHub CI pipelines where the earthly command is based on the image used for the outer container.

Hi, I started working on this issue, and came up with a minimal working implementation in https://github.com/GoogleCloudPlatform/spark-on-k8s-operator/pull/1296 My approach is to put a `template` property in SparkPodSpec, and serialize its...

@liyinan926 could you take a look?

why isn't this merged yet?

Also, frameless's current behavior is incompatible with Datasets ```scala import spark.implicits._ val a: Option[Int] = None val b: Option[Int] = None val data = X2(a, b) :: X2(a, a) ::...

Sorry, not within case classes, but when the option is parameterized over a case class.

From the user perspective I think the most natural mapping from `untyped.filter($"a" === null)` to frameless is `dataset.filter(A === None)`, which I would expect to give the same results. One...

If we want to retain the current, non-SQL compatible behavior, I think we should at least explicitly state in the documentation that `Option`s do not behave according to Kleene logic...

It breaks as well of course. This is no exception.

``` scala> case class A(a: Option[String]) defined class A scala> case class B(b: Option[String]) defined class B scala> val a = spark.createDataset(Seq(A(None))) a: org.apache.spark.sql.Dataset[A] = [a: string] scala> val b...