This is a counterpoint to Josh’s recent post: Why PostgreSQL Doesn’t Have Query Hints. I don’t really disagree, except that I think that there are many different definitions of “hints” floating around, leading to a lot of confusion. I could subtitle this post “More Terminology Confusion” after my previous entry.
So, let’s pick a reasonable definition: “hints are some mechanism to influence the SQL planner to choose a better plan”. Why did I choose that definition? Because it’s the actual use case. If a user encounters a bad plan, or an unstable plan, they need a way to get it to choose a better plan. There’s plenty of room to argue about the right way to do that and the wrong way, but almost every DBMS allows some form of hints. Including PostgreSQL.