Fix an incorrect error message regarding the size of usize and isize in cast_precision_loss.

[SLUCHABLUB]

When trying to cast a usize or isize using as to a type of potentially smaller size the cast_precision_loss would claim that the size of usize is "32 or 64 bits". It is now corrected to "16, 32, or 64 bits".

changelog: [cast_precision_loss]: fix the error messages claim regarding the size of usize and isize.

[rustbot]

r? @llogiq

rustbot has assigned @llogiq.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[samueltardieu]

Shouldn't we give a better error message? (and use "may cause" since on 16 bit usize platforms this is fine)

For example:

casting `usize` to `f32` may cause a loss of precision (`usize` can be as large as 64 bits wide depending of the target architecture, but `f32`'s mantissa is only 23 bits wide)

[SLUCHABLUB]

Yes, something like that would be even better. I will look over it.

@rustbot author

[rustbot]

Reminder, once the PR becomes ready for a review, use @rustbot ready.

[ada4a]

Friendly ping from triage: @SLUCHABLUB do you still plan to work on this?

[github-actions]

Lintcheck changes for ff77a3e

Lint	Added	Removed	Changed
clippy::cast_precision_loss	0	0	187

This comment will be updated if you push new changes

[SLUCHABLUB]

Yes, sorry.

@rustbot ready

[ada4a]

No need to apologize! I try to phrase these in a way that doesn't come off as indicting, but I don't always succeed..

[llogiq]

Thank you for both improving the message and simplifying the code.

View changes since this review

[llogiq]

This looks good, just needs a rebase.

@rustbot author

[rustbot]

This PR was rebased onto a different master commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

[SLUCHABLUB]

It seems that the 32-bit tests are failing. Is there a way for me to run (& bless) them on my 64 bit machine?

[llogiq]

Yes, you could cross compile to x86. But I don't think we actually need to incur that complexity here.

[SLUCHABLUB]

I think the 32-bit tests should pass now.

@rustbot ready