Add: orphan rule rationale. #1755
Conversation
TmSalviano
commented
Mar 11, 2025
•
TmSalviano
commented
- I thought it would be helpful to introduce the reader to the problem the orphan rule solves before being exposed to the rule itself. Understanding the orphan rule for the first time can be challenging, and knowing why it exists beforehand can make it easier to grasp.
- Also, including an explanation of the rationale behind the orphan rule in the orphan rule section of the page just makes sense.
rustbot
added
the
S-waiting-on-review
Co-authored-by: Josh Triplett <[email protected]>
| implementation is local to your crate, ensuring only one crate defines the implementation and | ||
| thereby maintaining coherence. |
There was a problem hiding this comment.
I would personally put a “that” here:
| implementation is local to your crate, ensuring only one crate defines the implementation and | |
| thereby maintaining coherence. | |
| implementation is local to your crate, ensuring that only one crate defines the implementation | |
| and thereby maintaining coherence. |
| r[items.impl.trait.orphan-rule.rationale] | ||
| The orphan rule helps ensure that other people's code can't break your code, and vice versa. | ||
| If an external crate implements an external trait for an external type, and your crate also | ||
| implements the same trait for the same type, the compiler wouldn't know which implementation | ||
| to use.\ | ||
| The orphan rule prevents this by requiring that either the trait or some type in the | ||
| implementation is local to your crate, ensuring only one crate defines the implementation and | ||
| thereby maintaining coherence. |
There was a problem hiding this comment.
Probably we'll want to move this into a note section, but it's generally going in the right direction. The one caveat I'd add is that the problem being prevented here isn't really about the compiler not knowing what to pick. The compiler knows what to do about that. It'd throw an ambiguity error. But what that means is that adding an impl for your own types or of your own traits would be breaking (in many more situations). So it's mostly about saving space for such authors, and that's the real rationale. Of course, whenever you talk about removing the rule, you do find out all the ways that we are leaning on the rule more technically, and those reasons are a bit more subtle. E.g.:
