Refactor ptr_offset_with_cast

[folkertdev]

Putting this lint in /methods is simpler and probably slightly more efficient.

changelog: [ptr_offset_with_cast]: respect MSRV when suggesting fix, and lint even more cases

[rustbot]

r? @Alexendoo

rustbot has assigned @Alexendoo.
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

[github-actions]

Lintcheck changes for f981739

Lint	Added	Removed	Changed
clippy::ptr_offset_with_cast	0	0	2

This comment will be updated if you push new changes

[folkertdev]

Hmm this time rustbot is mistaken, I'm moving a lint here, not adding a new one.

[ada4a]

Yeah that's a bit annoying..

Here's the incantation to restore the order:
@rustbot label -A-lint -S-blocked S-waiting-on-review
@rustbot note remove Feature-freeze

[samueltardieu]

Since this is a refactoring, some things might be made better as well.

For example:

• ptr.offset(…) is available since 1.0.0, while ptr.add(…) is available only since 1.26.0. A MSRV check is necessary in order not to suggest unapplicable fixes.
• Same happens with ptr.wrapping_offset(…) (1.16.0) and ptr.wrapping_add(…) (1.26.0).

Fortunately, all of those methods are const since the same version, so no special handling is necessary for const contexts.

Also, it would be preferable and more robust to use a multipart suggestion rather than building the new expression as a string:

• Get the method span and replace it with the new method name
• Compute the span of the as … part by truncating the cast expression span after the lhs and remove it

What do you think?

r? samueltardieu @rustbot author

View changes since this review

[rustbot]

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

[folkertdev]

That all sounds good to me.

@rustbot ready

[samueltardieu]

Looks good, except for the MSRV constant name which is too general. Also, could you add an example with &*const T so that we can ensure that this is properly detected and that the suggestion is sound?

View changes since this review

[samueltardieu]

I have edited the top-level comment to include a proper changelog as it will be included in the merge commit. For reference, the previous content was:

based on some comments by @ada4a in #15606. I copied the structure of this lint in that PR, so if that can be cleaned up, this can too.

Putting this lint in /methods is simpler and probably slightly more efficient.

changelog: none

[folkertdev]

Went with POINTER_ADD_SUB_METHODS, that should be clear and also covers the subtraction for later.

@rustbot ready

[samueltardieu]

It looks good to me, except for the test which is not isolated (even though it is nice to see that this works fine with the double fix!).

View changes since this review

[folkertdev]

@rustbot ready

[samueltardieu]

Thanks!

View changes since this review