@mhoye I'm interested in learning what we can do better; here, for example, is a review of 2 papers about bad compiler error messages, with recommendations, from 2018 and 2019. neverworkintheory.org/2021/09/

While it's good to see effort into making errors better (they have more context than docs so ought to be able to give good advice) it bothers me that these took till 2019 to appear.

(to be fair; the 2019 paper is a review of a long history of bad messages)