Michael J. Sullivan
Michael J. Sullivan
The index part of this issue is also just a bug, I think. I think we should disable compiling access policies when doing indexes, like we did for constraints in...
The tricky thing in this case is that the `id` that conflicts might be from a different type, and there are semantic questions about what should happen in an else...
The nice thing, though, is that `else` is typically not *needed*, it is just nice to have. So here, you can do ``` with id := $id, display_name := $display_name,...
I produced this mess of an implementation: ``` create or replace function json_merge(objs jsonb[], lo int8, hi int8) returns jsonb as $function$ select ( case when (lo + 1 =...
Yeah, final signature is `function json_object_pack(els: SET OF tuple) -> json`. Doing it as an aggregate function will mean needing to special case it in the compiler, which is fine....
Yeah, simple repro is trying to migrate to ``` type X { property tier -> enum; } ``` The error shows up in a different place if you use DDL...
We probably want to build a tool for it, then?
What is the correct behavior here? Rejecting the query because `target` is not a link attribute?
Ah, ok. And `SELECT Foo.link@target` ought to deduplicate, just like `SELECT Foo.link` does, I assume? (I ask because my first attempt on this was to directly patch the crashing code...
Hm, though actually maybe not deduplicating makes more sense, since it is operating on the *link*...