Open ilyanoskov opened 3 months ago
Defining each window as a separate table (with groupby) is probably a good idea. As for joining it all back, it is convenient to use ix_ref
- here is an example in the SQL-window spirit to compare the salary of an employee to the average salary in their position & department: https://pathway.com/developers/user-guide/data-transformation/indexing-grouped-tables/#multi-values-indexing.
Calling in @izulin - maybe you know of something cleaner. Either way, it would help to explain that this is the recommended way to implement SQL-windows (not to be confused with streaming windows).
Thanks a lot @dxtrous, will give it a try!
With Pathway, I found myself having to define each window as a separate table, and then joining them back together. The resulting code was quite verbose.
Unless I am missing something and it's possible to do it succinctly with Pathway?
Hi Ilya,
could you post an example of a verbose/unwieldy pathway code here? This could be a great starting point on how to extend syntax and/or tutorials.
I wonder if it's possible to have some additional syntax to make working with many windows possible?
For example, with ClickHouse SQL it's possible to work across many windows like this:
With Pathway, I found myself having to define each window as a separate table, and then joining them back together. The resulting code was quite verbose.
Unless I am missing something and it's possible to do it succinctly with Pathway?