Travis Staton

I'll also note two other desirable properties that we get from this waiter queue: 1. Fairer resource distribution as once a thread is enqueued in the waiter queue no later...

Does the "state hack" thwart the sharing attempt in this case? I don't see how yet: As I understand it, the "state hack" is to label all lambdas binding `State#`...

I believe that the original query was an example of the following problem: a lateral subquery that does not get pulled up constrains the planner to do a nested loop...

Interesting, it looks ilke in the `Tabulate` plan that the `left outer join lateral` doesn't contain an `array_agg` because that is pushed into another subquery within the lateral subquery, and...

I have this setup, so maybe I can offer some help before documentation is improved: You need to create the postgresql database that attic will connect to, and create a...

@ceedubs the expensive part of a diff is forcing the deep* fields in the cached root branch. In the transcript you posted the delete forces those thunks when it produces...

I expect the pattern match coverage checker to handle empty cases fine as is. I assume that error is from the parser, so maybe all that needs changing is the...

A match with no cases sounds good to me, and I expect the pattern match coverage checking code to work without modification. So, I think the parser just needs to...

Looks to me like this would be caused by the final "chunk delete" query's [where chunk.id in (POTENTIALLY_MASSIVE_LIST)](https://github.com/zhaofengli/attic/blob/4fedffe6a1020edfcfa7bef18d21321d4983b3a7/server/src/gc.rs#L231) condition. This could be solved by using a cursor or batching. In...