I would like to *educate* myself about why all *sql implicits* (most
notably conversion to Dataset API) are imported from *instance* of
SparkSession and not using static imports.
Having this design one runs into problems like this
It requires the presence of SparkSession instance (the only one we have) in
many parts of code. This makes code structuring harder.
I assume that there is a *reason* why this design was *chosen*. Can
somebody please point me to a resource or explain why is this?
What is an advantage of this approach?
Or why it is not possible to implement it with static imports?
Thanks a lot!