Ben Chambers

If I'm reading that correctly this is (as speculated on the PR) not a bug, but rather due to the fact that a `record` is never `null` (it's at-most a...

We almost centaily should not change records of all `null` fields to be `null`. This would be inconsistent with Python, SQL extensions supporting records, and a variety of other things....

At a high level, this is nice -- it makes it easier to work with Literals. We currently intend to distinguish between `add` and `add_scalar` instructions. This allows us to...

Elaborating a bit -- there seem to be two use cases: 1. I want the complete result of a windowed aggregate, once per window 2. I want the cumulative result...

I think we had a reason why we said it *was* the correct value... Wanted to add that this can be "hacked" for now: ``` sum(e, window=since(p)) count(e, window=since(p)) ```...

In notebooks, when wanting to emulate the behavior I want, I have previously used a filter to `null` out rows after the end: For example, for count: ``` let page_views_since_purchase_raw...

Which one is `x` in `x as i32`? I believe that one of those only supports casting to `i32`, but I could be wrong. It's weird how the error is...

Two thoughts: 1. I could see a case where the user wants to separate prepare and output, since they have different roles. Ideally this could be separate options. 2. Why...

Also probably eventually want an option for controlling where the snapshots (rocksdb) are written.

I think it depends on whether it is possible for the user to specify today (eg., by passing extra arguments to Wren via the session builder). If that's the case,...