closes #5
Finally I got some time to work on this task, I wanted to lay the foundation on how we build transformations for metrics, essentially I was inspired by how sqlite uses bytescodes instructions to generate prepare statements, and ended implementing basic solution following that pattern but I think it is flexible enough for us to generate all kind of built in transformations for metrics. i.e we can build a function called count_over_window(time_col,value_col,window_resolution) and behind the scene we build the require metrics by using already existing Instructions or create a new one if necessary.
Currently I focused the solution only on Datafusion later on we can refactor it to support more than one execution engine.
After this is merged I am going to create a set of issues to complete the instructions.
closes #5 Finally I got some time to work on this task, I wanted to lay the foundation on how we build transformations for metrics, essentially I was inspired by how sqlite uses bytescodes instructions to generate prepare statements, and ended implementing basic solution following that pattern but I think it is flexible enough for us to generate all kind of built in transformations for metrics. i.e we can build a function called count_over_window(time_col,value_col,window_resolution) and behind the scene we build the require metrics by using already existing Instructions or create a new one if necessary.
Currently I focused the solution only on Datafusion later on we can refactor it to support more than one execution engine.
After this is merged I am going to create a set of issues to complete the instructions.