Post

Asserting something is an `object` does not mean it becomes a type which has any properties (ie. `any`)

I feel the benefits are mostly in the signals: I use type errors when renaming or removing properties to know what code to update. If `obj?.a` is always valid, I lose those signals. New devs might use `?` as a crutch when property access fails. Typos like `obj?.a` vs `obj?.A` won’t be caught.

I think the reasoning here is that object does not have any keys, and you must have a declared type to avoid typos. You could just as easily have typed obj?.aa

oh my god why T_T

I had this "problem" just last week. TS only allow property access for known properties, even when it's optional. In other words, even though `null?.a` is valid, TS is saying `null` doesn't have a known property `a`.

On my phone so I can't check, but I think null (and some others?) are typeof object

It's because the object type doesn't have a property called a. Using `in` will probably solve that like you said 👀

Why not?

Ah, misunderstood you!

Curiously the error message is better when using v['a'] instead of v.a. Then it just straight out says "Property 'a' does not exist on type '{}'."

Hey while you’re fixing TypeScript can you get them to let me skip writing `Promise<>` around the return types of functions that already have the `async` keyword?

I offered it mostly as a stylistic alternative that, to me, feels more explicit. Not meant as a prescription.