hold up hold up so by trying to attribute a supply/withdraw event, vault v1 shows up in the morpho market events. But I need to track the relationship between v1 -> adapter (SCD2) -> v2 -> curator (SCD2) 😵‍💫

Interestingly, I always thought if you reference another Dune query, it would be pretty inefficient because the Trino optimizer doesn't plan the external query and the current query together. But it turns out it does. I have Query A: a supply/withdraw Morpho markets table. This is the full events log, so it should be quite heavy. Then Query B, that references Query A, filters on one specific market_id. It turns out Trino is still smart enough to do predicate pushdown (fancy term for filtering as early as possible) on Query A. Simply put, it transferred my market_id filter to Query A even though the filter was applied on Query B. I am not sure if, for more complex queries, Trino will be doing the same thing. But the implications of this are: You may not need to pre-optimise or do early filtering on the base tables. If your filters are on the last-mile tables, which are tables that are typically used to make dashboards, Trino is able to push down the filters upstream (what a strange sentence). I was pretty concerned about this at first and kind of over-optimized. But the word of warning is that you can only do this if you already expect that the downstream tables will have some sort of filtering. Because if you do not do this optimization at the upstream tables, it is going to cost you a lot of credits. Of course the standard rule applies in optimisation such as if you do a window function before filtering on particular values then it's going to cook you because you will be window functioning across the entire dataset. Not good. So really, the design of the query depends on the anticipated use case of the downstream table. Not sure if I'm making sense or if this is right. Maybe someone can also check this out. It's pretty cool though.

how tf is nescafe gold so expensive lately? Like its nearly $20 per 200g bottle