std::error::Error -> Trait Implementations: lifetimes consistency improvement

[WiktorPrzetacznik]

This cleans up std::error::Error trait implementations lifetime inconsistency ('static -> 'a)

Reasoning:

Trait implementations for std::error::Error, like:
impl From<&str> for Box<dyn Error + 'static, Global>
impl<'a> From<&str> for Box<dyn Error + Sync + Send + 'a, Global>
use different lifetime annotations misleadingly implying using different life annotations here is a conscious, nonaccidental decision.

(Related forum discussion here)

[rustbot]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @joshtriplett (or someone else) soon.

Please see the contribution instructions for more information. Namely, in order to ensure the minimum review times lag, PR authors and assigned reviewers should ensure that the review label (S-waiting-on-review and S-waiting-on-author) stays updated, invoking these commands when appropriate:

• @rustbot author: the review is finished, PR author should check the comments and take action accordingly
• @rustbot review: the author is ready for a review, this PR will be queued again in the reviewer's queue

[workingjubilee]

@WiktorPrzetacznik Can you edit your post to include the reasoning to accept this PR instead of a link to a forum comment so I can kick this over to an FCP and that reasoning can be easily read and discussed by everyone instead of having to comb through that thread for context (or even to determine how much of that context is applicable)?

Note that just copying the comment (with attribution) should be fine.

[WiktorPrzetacznik]

@workingjubilee Thank you for the suggestion, reasoning is now included.

[workingjubilee]

Thanks!

r? libs-api Requesting FCP for relaxing the lifetime constraints of these implementations for trait objects where they currently can imply the 'static lifetime in a few cases instead of 'a due to trait object lifetime defaults.

[dtolnay]

@rust-lang/libs-api:
@rfcbot fcp merge

This fixes the spurious lifetime error in code such as the following.

fn main() {
    let non_static_error = std::fmt::Error;
    let mut boxed_error: Box<dyn std::error::Error> = Box::new(&non_static_error);
    replace_error(&mut boxed_error);
}

fn replace_error(out: &mut impl From<String>) {
    *out = "...".to_owned().into();
}

error[E0597]: `non_static_error` does not live long enough
 --> src/main.rs:3:64
  |
2 |     let non_static_error = std::fmt::Error;
  |         ---------------- binding `non_static_error` declared here
3 |     let mut boxed_error: Box<dyn std::error::Error> = Box::new(&non_static_error);
  |                                                       ---------^^^^^^^^^^^^^^^^^-
  |                                                       |        |
  |                                                       |        borrowed value does not live long enough
  |                                                       cast requires that `non_static_error` is borrowed for `'static`
4 |     replace_error(&mut boxed_error);
5 | }
  | - `non_static_error` dropped here while still borrowed

[rfcbot]

Team member @dtolnay has proposed to merge this. The next step is review by the rest of the tagged team members:

• @Amanieu
• @BurntSushi
• @dtolnay
• @joshtriplett
• @m-ou-se

Concerns:

• crater-run-first resolved by std::error::Error -> Trait Implementations: lifetimes consistency improvement #113833 (comment)

Once a majority of reviewers approve (and at most 2 approvals are outstanding), this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

See this document for info about what commands tagged team members can give me.

[BurntSushi]

Can we do a crater run for this change? It seems unlikely it will break anyone, but I think it would be good to check. I'm thinking in particular about regressions related to inference failures

[BurntSushi]

@rfcbot concern crater-run-first

[workingjubilee]

@bors try

[dtolnay]

4 crates regressed with a "build failure" and all 4 look spurious to me. Out of memory, disk full, some kind of container disconnect. None of these could possibly be related to a compile error from the lifetimes change. @BurntSushi could you take a look as well, and resolve the concern?

[BurntSushi]

@rfcbot resolve crater-run-first

[rfcbot]

🔔 This is now entering its final comment period, as per the review above. 🔔

[rfcbot]

The final comment period, with a disposition to merge, as per the review above, is now complete.

As the automated representative of the governance process, I would like to thank the author for their work and everyone else who contributed.

This will be merged soon.

[dtolnay]

Thanks!

[dtolnay]

@bors r+

[bors]

📌 Commit 1da87ee has been approved by dtolnay

It is now in the queue for this repository.

[bors]

🌲 The tree is currently closed for pull requests below priority 100. This pull request will be tested once the tree is reopened.