Maximilian Roos

This would be great. We probably have to add it into the compiler (can't be done in stdlib); either specific to the `in` function, or to compile a PRQL list...

> I'm wondering whether the following query might not solve the problem as stated and be the simplest and most idiomatic PRQL? > > ```elm > from foo > derive...

> On a perhaps (un)related note, I'm not a fan of the implicit window functions. I was surprised when I became of them in another thread with aljazerzen. Yeah, I...

> Perhaps the current `in` function may need to be renamed? I think ideally we should support both usages, with the same name...

Briefly on one point — and sorry for not having responded above > That's not true: when using LAG, you want the whole window but also want to have order....

> There are two types of windowing: > > * actual window functions that produce the whole column at once (LAG, RANK, NTILE, ...). They map n rows into n...

My message above is wrong — I was thinking of ranking one column (e.g. `score`) in a rolling window of another (e.g. `date`). SQL doesn't do that. Not sure if...

So to confirm my SQL understanding, it seems there are three categories of window / "analytic" functions: - Aggregate analytic functions — sum, count, etc — aggregate functions with a...

If that's the case, then I can see a few options for our defaults: 1. Inherit SQL defaults - Everything works now. The defaults are arguably normally the thing folks...

(I realize my numbers didn't agree with my titles, so fixed above) Full window was also my vote; do you think it's viable to also let dialects like BQ work...