Harlan Haskins in Elk: "I think the fact that Task closures ..."

I think the fact that Task closures don't require explicit self is a huge footgun :(

December 11, 2025 at 4:17:37 PM

is it? Isn’t it like UIView.animate, where its generally regarded safe to strongly capture self

Not if you save the Task to cancel it later, which is the case in my experience for most long-running tasks that await over async sequences

Agree! Also allowing you to call throwing functions without a do/catch block is really bad

Yes! It took me a bunch of debugging to find this out the hard way.

I made a tiny utility library to help with this: https://swiftpackageindex.com/samsonjs/AsyncMonitor

I agree

Having closures w/o a GC is the real footgun

Explicit self is just one special case that doesn’t deserve an own solution. The problem affects any capture, as each capture might also refer to self or produce other cycles.
I think what you really want is explicit capture lists, not explicit self in the body. (the { [self, a, b] in … } thing)

Replying to

yes, but for the unrelated reason that I just don’t like implicit self ducks

I don’t think that is unrelated and also I agree!