Move format_push_string to restriction

[Victor-N-Suadicani]

Fixes #9077 (kinda) by moving the lint to the restriction group. As I noted in that issue, I think the suggested change is too much and as the OP of the issue points out, the ramifications of the change are not necessarily easily understood. As such I don't think the lint should be enabled by default.

changelog: [format_push_string]: moved to restriction (see #9077).

[rust-highfive]

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

Please see the contribution instructions for more information.

[yoav-lavi]

Hey, author of #9077.

Thank you!

If we want to move the lint, I'd like to note that using clippy::pedantic is (anecdotally) popular from what I've seen, so perhaps clippy::restrictions (or clippy::nursery)?

Considering that the issues with the lint don't really fall into "occasionally a false positive" or "too strict", I personally think it falls more into the category of a restriction

(Although to be fair, I use clippy::pedantic so I may be biased 🙂)

[Victor-N-Suadicani]

I totally forgot that restriction even existed and I agree that pedantic wasn't the best fit. I've moved it to restriction now :)

[flip1995]

Thanks! I agree with the reasoning in the linked issue and that it should be moved to restriction. Can you please add a summary of this issue to a new ### Known Issues section in the link documentation after the why is this bad section? This should just summarize, that changing the code to write will lead to usage of a fallible API.

[Jarcho]

Just going to add a couple points here.

• format! can fail for exactly the same reason write! to a string can fail. It just calls expect internally.
• The only way this can return an error is if a Display implementation creates the error. This would be a violation of the stated API

I see this as a combination of the lint docs not being enough to justify the change, and an unfortunate side-effect of having to make the unwrap call explicit.

[yoav-lavi]

@Jarcho

format! can fail for exactly the same reason write! to a string can fail. It just calls expect internally.

That was my assumption, the lint (in my opinion) doesn't make that clear enough (with the added issue of encouraging ignoring an error ad @Victor-N-Suadicani stated in the issue)

[Victor-N-Suadicani]

@flip1995 Added the known problems section :)

[flip1995]

@bors r+

Thanks!

[bors]

📌 Commit 43ac256 has been approved by flip1995

It is now in the queue for this repository.

[bors]

⌛ Testing commit 43ac256 with merge 10e85ef...

[bors]

☀️ Test successful - checks-action_dev_test, checks-action_remark_test, checks-action_test
Approved by: flip1995
Pushing 10e85ef to master...