this post was submitted on 11 Jul 2023
8 points (100.0% liked)
Rust
5981 readers
50 users here now
Welcome to the Rust community! This is a place to discuss about the Rust programming language.
Wormhole
Credits
- The icon is a modified version of the official rust logo (changing the colors to a gradient and black background)
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Rust has a strong emphasis on strongly typing constraints. So if a collection of fields are always some/none together having them listed as separate options in the struct means there is some assumptions you are making that the type system isn't aware of which can lead to pain and bugs in the future.
I agree that separate types with
Into
sounds like a nice solution to me, it would be good to see the error the compiler is giving you (or if you can a minimal reproducible repo). If you absolutely can't make it work, a singleOption<Inner>
at least would be more correct as all the fields on the inner struct would be optional together.Wait really? If I wrap a struct in Option it makes all the fields optional?
Good to know that you think the Into approach seems better. Part of the purpose of this thread is to just gauge what's the better way to do this in Rust. Do you know what "separating the types with Into" would look like?
Yeah, sort of. I probably didn't explain super well, and also probably don't fully understand the problem so here are some code snippets that might make things more concrete and you can tell me where my asumptions of your codebase are wrong
So first off we have what I assume you were suggesting with multiple options for the individual db props. I commented where things are painful and bug prone:
https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=2ecb08dd5d54378c693a7014599e8645
If
Option
is indeed the aproach you want to take we can solve a lot of these problems by moving all the fields that go optional together into a separate struct:https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=9eb2f896e171507acb64283d5b530673
We get even more type safety and clarity by making them separate types (and have ArmyWithDbProps wrap Army):
https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=c2e33aa1c09538ffdbe0cd440bbff3e1
Or we could use Into if it's not appropriate for
process_army_from_db
to turn anArmyWithDbProps
into anArmy
:https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=f82043935625572cb08c240f23944f0e
Also this last example is using clone when if this is in-fact the direction you want to go we could be using pointers to avoid unnecessary clones. Let me know if this is the case and I can write a version with pointers and lifetimes.
I'm sure we've missed something here specific to your software (obviously the above links are all trivial examples), but I just wanted to help clarify my original point with some concrete code. If you can share some of your code we might be able to give you more specific advice.
I'd provide more code but it's a mess and full of old commented-out code. Your examples are perfect! combining the DB fields into it's own struct is something I hadn't thought of... and I totally get why having a bunch of options sitting in the Army struct would be problematic. I'm really excited about rust for moving these sorts of errors to compile time.
The INTO example seems great too. I'm ok with the performance hit of cloning for now... lifetimes and pointers feel like a tier above where am at with my rust skills, and I'll circle back to get a better handle on them later.
One question about the INTO example... I always hear it's better to just implement FROM and get INTO for free. Does that not make sense for my use case? If I did it, would it look something like:
impl From<ArmyWithDbProps> for Army { fn from(self) -> ArmyWithDbProps { self.armyWithDbProps } }