News for hackers. Stuff that matters.
Re:PRQL a simple, powerful, pipelined SQL replacement by Nihilartikel on Friday December 30, 2022 @09:13AM (#34185143)
That's what I want too.

To make a loose metaphor, if SQL is like high level opengl, then I want the low level Vulkan and shader APIs for hand optimization.

I would like this especially for Apache Spark sql, since a bad plan that isn't aware of the data shape can blow gigabytes over the network unnecessarily.

Re:PRQL a simple, powerful, pipelined SQL replacement by tormeh on Friday December 30, 2022 @08:14PM (#34192208)
Spark has datasets, which I think lets you do this, although it’s only supported in Java and Scala, not Python.