Zach Daniel

I still think it's more complicated than just deciding what is/isn't mergeable, but that's besides the point. Attributes with exact matching types is good enough for now.

🤔 Honestly we do very little compile time optimization. What I'm getting at with my comments is that we're talking about introducing a generic `Spec` concept, designed to ensure that...

Strange! I can't reproduce this in spark even.

Where in the module are you aliasing? It shouldn't matter, but I can't reproduce and maybe somehow that matters

We would synthesize the indices based on batch sizes, so we won't have to return records

Like add the next 500 integers to the stream.

🤔 I was thinking that `emit_indices` returns the number of inputs that were handled, not the number of records that were created.

🤔 potentially. It could be confusing the other way around as well, like if you do huge bulk upsert and everything is match and you get back `0`. But maybe...

Yes, it's currently always over a declared relationship. We have plans to support using a resource instead of a relationship path for much more custom aggregation. https://github.com/ash-project/ash/issues/939

Right, so the behavior that needs to be fixed here is a couple things. 1. as you mentioned, `return_notifications?` option must be added & honored. 2. additionally, when you are...