[jmull]

I actually think this wouldn’t be that helpful.

The developer still needs to ensure they apply their set of pragmas, whether that’s a single edition pragma or a set of pragmas. And they still need to understand and carefully choose the pragmas/options they use (an edition really makes this a little harder by abstracting/hiding something that needs to be directly understood and visible).

And these proposed new default pragmas are more incremental improvements rather than complete solutions, more convenience than critical. E.g., while the default affinity is goofy, strict tables don’t come close to a comprehensive validation mechanism — so if you need strong validation you’re likely going to need to implement that at a higher level anyway. (Also, there’s a decent separation-of-concerns argument that you should handle it separately.)

Likewise the busy timeout. The pragma is convenient but you still need to handle busy timeouts. (The author’s problem, “I didn’t realize busy timeouts could happen so the app didn’t handle them correctly”, is not solved by the pragma.)