Implement slice_strip feature

[tesuji]

Tracking issue: #73413

[Mark-Simulacrum]

Does making the impls here heterogeneous (i.e. impl<T> [T] { fn strip_prefix<U>(&self, prefix: &[U]); } hurt inference sufficiently that we don't want to do that?

[jonas-schievink]

Unsure why I'm assigned, I'm not on the libs team. r? @Mark-Simulacrum

[tesuji]

hurt inference sufficiently

I don't understand the question.
Why would we want this?

impl<T> [T] { 
    fn strip_prefix<U>(&self, prefix: &[U]); 
}

[tesuji]

Also, the strip_* methods somewhat copy the signature of slice::starts_with.

Does starts_with also cause problems?

pub fn starts_with(&self, needle: &[T]) -> bool

[tesuji]

r? @SimonSapin

[tesuji]

r? @dtolnay

[dtolnay]

Thanks! I am on board with this. It matches the method names on str and it matches the starts/ends_with signatures on slice.

[dtolnay]

@bors r+

[bors]

📌 Commit 560a996 has been approved by dtolnay

[Manishearth]

test time.rs - time::Duration::zero (line 145) ... ok
test unit.rs - unit::() (line 8) ... ok

failures:

---- slice/mod.rs - slice::[T]::strip_prefix (line 1468) stdout ----
error[E0308]: mismatched types
---- slice/mod.rs - slice::[T]::strip_suffix (line 1519) stdout ----
error[E0308]: mismatched types
 --> slice/mod.rs:1522:1
  |
6 | assert_eq!(v.strip_suffix(&[]), Some(v));
  | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ expected slice `[{integer}]`, found array `[{integer}; 3]`
  |
  = note: expected enum `std::option::Option<&[{integer}]>`
             found enum `std::option::Option<&[{integer}; 3]>`
  = note: this error originates in a macro (in Nightly builds, run with -Z macro-backtrace for more info)

error: aborting due to previous error

For more information about this error, try `rustc --explain E0308`.
Couldn't compile the test.

failures:
    slice/mod.rs - slice::[T]::strip_prefix (line 1468)
    slice/mod.rs - slice::[T]::strip_prefix (line 1479)
    slice/mod.rs - slice::[T]::strip_suffix (line 1508)
    slice/mod.rs - slice::[T]::strip_suffix (line 1519)

test result: FAILED. 2600 passed; 4 failed; 28 ignored; 0 measured; 0 filtered out

#73733 (comment)

@bors r-

[tesuji]

Is this example needed?

[tesuji]

I removed it.

[dtolnay]

I think the wording of this was clearer in a previous revision. Slice not starting with prefix is a scenario that is important to specify; this says what the meaning of Option in the return type is. Empty prefix is a small clarification (not even an edge case, since the behavior falls out from the rest of the documentation). I think the emphasis in this paragraph is backward and not conducive to recognizing the part that is more important.

[tesuji]

I repharsed it. Could you take a look?

[dtolnay]

@bors r+

[bors]

📌 Commit cd9d833 has been approved by dtolnay