The article title is clickbaity as heck, but the content is (reasonably) sound.
A better title would be 'Rust's async model has some flaws that are a product of its desire to ensure safety and make the true cost of certain operations visible'.
Of course, such a title would never hit the front page of hacker news.
The problems the author identifies are real, but they're not unique to Rust. You can get around them fairly easily with dynamic dispatch, which is what basically every competitor to Rust's async model does by default.
I agree with the author that Rust's use of async in the ecosystem is increasingly universal and it's starting to become a little problematic. As the author of Flume and a (soon to be released) concurrent hashmap implementation, I'm trying to work on adding primitives that successfully marry sync and async with one-another. There's no doubt that it's hard though, and function colouring is a problem.
A better title would be 'Rust's async model has some flaws that are a product of its desire to ensure safety and make the true cost of certain operations visible'.
Of course, such a title would never hit the front page of hacker news.
The problems the author identifies are real, but they're not unique to Rust. You can get around them fairly easily with dynamic dispatch, which is what basically every competitor to Rust's async model does by default.
I agree with the author that Rust's use of async in the ecosystem is increasingly universal and it's starting to become a little problematic. As the author of Flume and a (soon to be released) concurrent hashmap implementation, I'm trying to work on adding primitives that successfully marry sync and async with one-another. There's no doubt that it's hard though, and function colouring is a problem.