This is amazing. We moved from Oracle to Postgres after being on Oracle for 15+ years. That was a huge move that took 2 years. One of the biggest hurdles was rewriting thousands of sql queries. One of the biggest things we had to change was accounting for how oracle handled nulls versus how Postgres handled nulls. However, we missed this nuance about how Postgres treated nulls and unique. It wasn’t until about 2 years into using Postgres, and running into so many issues, that we finally stumbled upon this nuance in the documentation. Crap. That required making a bunch of more changes to account for this. At this point I’m firmly convinced that null handling is the hardest part of database work ;-)