pgdog looks interesting, but I read the docs wondering how it handles foreign keys between tables, and the docs don't seem to cover it. It was the first question I had, and I'd assume would be an important question for most users. The project states "Data relationships are not sacrificed to go faster" and that it operates "transparently to the application", but it's not clear how it does that in regard to foreign keys.

Additionally, maybe this is just a pet peeve with Postgres documentation in general, but the docs use "database" to mean both a single database and a Postgres server interchangeably. On some pages this makes the content harder to understand. I think it's good practice to distinguish between databases and "clusters" (no, not those clusters, actually servers), and perhaps between clusters and groups of servers. While some of the naming is unfortunate for historical reasons, re-using the same names for different concepts just causes more confusion.