Automatically deriving Clone is a convenience. You can and should write your own implementation for Clone whenever you need it and automatically derived implementation is insufficient.
replies(3):
By the way, this issue also affects all of the other derivable traits in std - including PartialEq, Debug and others. Manually deriving all this stuff - especially Debug - is needless pain. Especially as your structs change and you need to (or forget to) maintain all this stuff.
Elegant software is measured in the number of lines of code you didn't need to write.
Nevertheless, it would be cool to be able to add #[noderive(Trait)] or something to a field not to be included in automatic trait implementation. Especially that sometimes foreign types do not implement some traits and one has to implement lots of boilerplate just to ignore fields of those types.
I know of Derivative crate [1], but it's yet another dependency in an increasingly NPM-like dependency tree of a modern Rust project.
All in all, I resort to manual trait implementations when needed, just as GP.
I rather have a simple and explicit language with a bit more typing, than a perl that tries to include 10.000 convenience hacks.
(Something like Uiua is ok too, but their tacitness comes from simplicity not convenience.)
Debug is a great example for this. Is derived debug convenient? Sure. Does it produce good error messages? No. How could it? Only you know what fields are important and how they should be presented. (maybe convert the binary fields to hex, or display the bitset as a bit matrix)
We're leaving so much elegance and beauty in software engineering on the table, just because we're lazy.
Strong disagree. Elegant software is easy to understand when read, without extraneous design elements, but can easily have greater or fewer lines of code than an inelegant solution.
[1] https://rustsec.org/advisories/RUSTSEC-2024-0388.html
[2] https://crates.io/crates/derive_more
Unfortunately this problem only shows up when you’re combining derive with certain generic parameters for the first time. The first time I saw this, I thought the mistake was mine. It was so surprising and confusing that it took half an hour to figure out what the problem was. I thought it was a compiler bug for awhile and went to file it on the rust project - only to find lots of people had beat me to it.
Aside from anything else, it’d be great if rust had better error messages when you run into this issue.
I wouldn't write anything but Rust via LLMs, because that's the only language where I feel that the static type checking story is strong enough, in large parts thanks to kani.
OpenAIs Codex model is also ridiculously capable compared to everything else, which helps a lot.
This is a pet peeve of mine so I'm sorry to be overly dismissive. There are bad error messages out there in the world that are impossible to parse, but this is not one of them. Trying to file a github issue before attempting to understand the error message is insane to me.
Would you mind filing a ticket detailing what you'd wish the error had been? Without additional context, the only improvement I can think of is adding a note explaining imperfect derives when hitting a missing trait bound coming from a local crate derived impl.
Take a look. Do you think this quirk of derive is obvious from the error message alone? Would you have figured it out, in the context of a much more complex program with dozens of other changes that may or may not have been related?
https://play.rust-lang.org/?version=stable&mode=debug&editio...
The compiler error says my type didn’t implement clone. But #[derive(Clone)] was right there on my type, and my type was obviously cloneable. The error wasn’t on the derive - or anywhere nearby. My program even compiled successfully, right up until I tried to actually call clone() on my object. And my type trivially, obviously was cloneable because all the fields were clonable.
My first thought at the time was that my derive attribute was being ignored by the compiler somehow, but it wasn’t at all obvious why. The compiler’s suggested fix was also wrong and misleading. It suggests adding clone to an irrelevant type that shouldn’t impl clone in the first place. That’s the wrong fix.
In summary, the error message doesn’t offer the real problem, which is that the trait bounds added by derive Clone were wrong. And it didn’t suggest the proper fix - which was to impl clone manually.
I was very confused for a good while with this one. Get pissy if you want but I find this incredibly counterintuitive and I found the compiler’s error message to be uncharacteristically unhelpful.
If this quirk of derive was truly obvious, this blog post wouldn’t have hit the front page of HN in the first place. I cut my hand on one of rust’s sharp edges. Don’t get mad at me for bleeding a little.
Something like this would have helped me immensely:
> Note: even though struct Foo has derive(Clone), Foo does not implement clone in this case. derive(Clone) may have overly restrictive trait bounds (impl Clone where T: Clone). If this is the case, you may need to manually implement Clone for Foo with less restrictive trait bounds:
impl Clone for Foo {
fn clone(&self) …My pet peeve is the learned helplessness around compiler error messages, particularly ones that go to great lengths to be informative and offer solutions instead of just throwing garbage at you.
I think software - like mathematics - becomes more elegant & beautiful when we find ways to simplify it without making it worse in other ways. Minimising the number of lines of code isn’t the goal. But sometimes making something smaller also makes it easier to understand and sometimes even more powerful at the same time. Those are the days I live for.
Alan Kay says the right point of view is worth 50 IQ points. There are plenty of examples of this in software. Innovations in ways of structuring our programs that make them at once simpler and more beautiful. All without sacrificing anything important in the process. We take them for granted now, but innovation is rarely obvious from the outset. Compilers and high level languages. Sum types. Operating systems. Maybe SQL and ECS. Sinatra’s http handler API. And so on.
Everyone said the same thing of most compiler errors until clang came along and showed everyone how much better compiler messages could be.
The compiler message here doesn’t suggest the true problem here, nor does it suggest the appropriate fix. I’m also clearly not the only one who finds this issue surprising and annoying. This blog post wouldn’t have been upvoted and commented on if it didn’t strike a nerve.
If nothing else, I think it’s quite obvious that the compiler’s error message could be improved. But better yet, I’d love to see it fixed in the language. The bounds derive places on its traits are simply wrong. There’s no reason why T needs to impl Clone. T is irrelevant.
For what it’s worth, I get what you’re frustrated about. I spent 2 years as a programming teacher. I feel like the first month of every cohort I wander around telling upset students that they forgot a semicolon. And the third month I wander around guiding frustrated students to read the compiler error message, which would have saved them 2 hours of guessing.
After banging my head into a wall for awhile on this issue, I made a reduced reproducing test case. Then I went looking in the rust compiler issue tracker. Only there did I finally see a proper explanation of what was going on, amidst a sea of other people struggling with the same thing.
That was half an hour of my life I won’t get back. I’m not helpless. But I am still pretty frustrated by the experience. I see it as a language level bug. It seems to take years and years of bike shedding for rust to fix stuff like this. I’d give even odds this still isn’t fixed in a decade from today.
But maybe, at least, we might be able to improve the error message?
The place this approach falls down for me is in refactoring. Sure, you can get chatgpt to help you write a big program. But when I work like that, I don’t have the sort of insights needed to simplify the program and make it more elegant. Or if I missed some crucial feature that requires some of the code to be rewritten, chatgpt isn’t anywhere near as helpful. Especially if I don’t understand the code as well as I would have if I authored it from scratch myself.