4 ms·

What you're describing sounds to me like higher kinded types, which are of some relationship to GATs (GATs enable a better encoding of higher kinded types as I

What you're describing sounds to me like higher kinded types, which are of some relationship to GATs (GATs enable a better encoding of higher kinded types as I understand it), but are not GATs. https://docs.rs/higher/latest/higher/ https://docs.rs/higher/latest/higher/

gunapologist99 4y ago
> as I understand it You could be 100% right (I have no idea, but it sounds reasonable), but this entire question really hits home that Rust has tons of basically incomprehensible design features. The worst part about features like this are that a few smart people will actually use them (whether because they're powerful features, or just because they feel powerful when using them), and then render their code incomprehensible to the mere mortals who have to support their code in the future. The KISS principle (https://en.wikipedia.org//wiki/KISS_principle https://en.wikipedia.org//wiki/KISS_principle) is sometimes ignored because people don't want to be stupid, and why wouldn't you want to use a new toy(tool) once you've figured it out?
- kibwen 4y ago
  This is quite uncharitable. Rust's features aren't there for no reason, they're there to address the hard problems of how to achieve C++-level performance while providing the level of abstraction that modern developers expect and remaining totally type-safe. Furthermore, the amount of hand-wringing over this feature is disproportionate to its actual impact. It feels like people see "Haskell" mentioned in the same breath and immediately lose all sense of reason. This feature is not basically incomprehensible, it's a straightforward addition that allows generics to be used with associated types, both of which are features that already exist in the language. If you're asking yourself, "when would I ever need to use these?", this answer is that you may never find call to use these. Traits are mechanisms for creating abstractions, which are mostly relevant to library authors. If you're not a library author, you may never have a good reason to write a trait. Even for people who are writing traits, they may have no reason to ever use associated types. And even for people who are writing traits with associated types, they may have no reason to ever want to use generics with those associated types. However, all of these features are extremely useful for the people consuming libraries, and far from making Rust code more complicated, from what I've seen so far it makes Rust code in the wild less complicated.
- soggybutter 4y ago
  Things like GATs also allow those smart people to add sorely needed features, like async traits, which the rest of us plebeians can benefit greatly from
- philzook 4y ago
  The problem is people who are smart enough to use fancy features, but not wise enough to show restraint. The question is do we inhibit the wise to protect us from the unwise? The answer might be yes. I enjoy learning about power features so I hope not.
- idontpost 4y ago
  > The worst part about features like this are that a few smart people will actually use them (whether because they're powerful features, or just because they feel powerful when using them), and then render their code incomprehensible to the mere mortals who have to support their code in the future. I really don't understand what motivates this kind of petty hostility. It could have been written about any of a thousand language features that didn't exist in assembly and now exist in higher level languages, and it would be just as sad.

gunapologist99 4y ago

> as I understand it You could be 100% right (I have no idea, but it sounds reasonable), but this entire question really hits home that Rust has tons of basically incomprehensible design features. The worst part about features like this are that a few smart people will actually use them (whether because they're powerful features, or just because they feel powerful when using them), and then render their code incomprehensible to the mere mortals who have to support their code in the future. The KISS principle (https://en.wikipedia.org//wiki/KISS_principle https://en.wikipedia.org//wiki/KISS_principle) is sometimes ignored because people don't want to be stupid, and why wouldn't you want to use a new toy(tool) once you've figured it out?

kibwen 4y ago
This is quite uncharitable. Rust's features aren't there for no reason, they're there to address the hard problems of how to achieve C++-level performance while providing the level of abstraction that modern developers expect and remaining totally type-safe. Furthermore, the amount of hand-wringing over this feature is disproportionate to its actual impact. It feels like people see "Haskell" mentioned in the same breath and immediately lose all sense of reason. This feature is not basically incomprehensible, it's a straightforward addition that allows generics to be used with associated types, both of which are features that already exist in the language. If you're asking yourself, "when would I ever need to use these?", this answer is that you may never find call to use these. Traits are mechanisms for creating abstractions, which are mostly relevant to library authors. If you're not a library author, you may never have a good reason to write a trait. Even for people who are writing traits, they may have no reason to ever use associated types. And even for people who are writing traits with associated types, they may have no reason to ever want to use generics with those associated types. However, all of these features are extremely useful for the people consuming libraries, and far from making Rust code more complicated, from what I've seen so far it makes Rust code in the wild less complicated.
soggybutter 4y ago
Things like GATs also allow those smart people to add sorely needed features, like async traits, which the rest of us plebeians can benefit greatly from
philzook 4y ago
The problem is people who are smart enough to use fancy features, but not wise enough to show restraint. The question is do we inhibit the wise to protect us from the unwise? The answer might be yes. I enjoy learning about power features so I hope not.
idontpost 4y ago
> The worst part about features like this are that a few smart people will actually use them (whether because they're powerful features, or just because they feel powerful when using them), and then render their code incomprehensible to the mere mortals who have to support their code in the future. I really don't understand what motivates this kind of petty hostility. It could have been written about any of a thousand language features that didn't exist in assembly and now exist in higher level languages, and it would be just as sad.