Can you make "somestr".trim_matches(char::is_ascii_punctuation) compile too ? it compiles with char::is_numeric

this works:

assert_eq!("123foo1bar123".trim_matches(char::is_numeric), "foo1bar"); //ok

but this doesn't:

assert_eq!(
        ".,\"foo1bar\".,';".trim_matches(char::is_ascii_punctuation),
        "foo1bar"
    ); //XXX fail
    // expected signature of `fn(char) -> _`
    // found signature of `for<'r> fn(&'r char) -> _`
    //   = note: required because of the requirements on the impl of `std::str::pattern::Pattern<'_>` for `for<'r> fn(&'r char) -> bool {std::char::methods::<impl char>::is_ascii_punctuation}`

sig for is_numeric is:

pub fn is_numeric(self) -> bool

And for is_ascii_punctuation is:

pub fn is_ascii_punctuation(&self) -> bool

So in order to make that work, I've had to do this:

pub trait Man {
    fn manual_is_ascii_punctuation(self) -> bool;
}

impl Man for char {
#[inline]
    fn manual_is_ascii_punctuation(self) -> bool {
        self.is_ascii() && (self as u8).is_ascii_punctuation()
    }
}
fn main() {
assert_eq!(
        ".,\"foo1bar\".,';".trim_matches(char::manual_is_ascii_punctuation),
        "foo1bar"
    ); //works because the func sig matches
}

Full code (playground):

#![allow(unused)]

pub trait Man {
    fn manual_is_ascii_punctuation(self) -> bool;
}

impl Man for char {
#[inline]
    fn manual_is_ascii_punctuation(self) -> bool {
        self.is_ascii() && (self as u8).is_ascii_punctuation()
    }
}

fn main() {
    assert_eq!("11foo1bar11".trim_matches('1'), "foo1bar");
    assert_eq!("123foo1bar123".trim_matches(char::is_numeric), "foo1bar"); //ok
    assert_eq!(
        ".,\"foo1bar\".,';".trim_matches(char::manual_is_ascii_punctuation),
        "foo1bar"
    ); //works because the func sig matches
    assert_eq!(
        ".,\"foo1bar\".,';".trim_matches(char::is_ascii_punctuation),
        "foo1bar"
    ); //XXX fail
    // expected signature of `fn(char) -> _`
    // found signature of `for<'r> fn(&'r char) -> _`
    //   = note: required because of the requirements on the impl of `std::str::pattern::Pattern<'_>` for `for<'r> fn(&'r char) -> bool {std::char::methods::<impl char>::is_ascii_punctuation}`

    assert_eq!("\"123foo1bar\"".trim_matches(|x| x == '"'), "123foo1bar"); //ok

    let x: &[_] = &['1', '2'];
    assert_eq!("12foo1bar12".trim_matches(x), "foo1bar");
}

Compiling playground v0.0.1 (/playground)
error[E0631]: type mismatch in function arguments
  --> src/main.rs:22:29
   |
22 |         ".,\"foo1bar\".,';".trim_matches(char::is_ascii_punctuation),
   |                             ^^^^^^^^^^^^
   |                             |
   |                             expected signature of `fn(char) -> _`
   |                             found signature of `for<'r> fn(&'r char) -> _`
   |
   = note: required because of the requirements on the impl of `std::str::pattern::Pattern<'_>` for `for<'r> fn(&'r char) -> bool {std::char::methods::<impl char>::is_ascii_punctuation}`

error: aborting due to previous error

For more information about this error, try `rustc --explain E0631`.
error: Could not compile `playground`.

To learn more, run the command again with --verbose.

[varkor]

So in order to make that work, I've had to do this:

You can just use a closure instead:

assert_eq!(
        ".,\"foo1bar\".,';".trim_matches(|c| char::is_ascii_punctuation(&c)),
        "foo1bar"
    );

[ExpHP]

Ouch. Looks like this is an unfortunate holdover from when these methods were a part of std::ascii::AsciiExt.

Technically speaking, I believe the answer is yes; this is technically possible. However, it cannot be done by implementing Pattern for F: FnMut(&char) -> bool, because that would overlap with the impl for F: FnMut(char) -> bool. There would have to be two impls that differ in a very subtle way:

// This would work for any closure.
impl<'a, F> Pattern<'a> for F where F: FnMut(char) -> bool { ... }

// This would only work for path expressions to fn items
//    and type-annotated locals.
// (even annotated closures like `|c: &char| true` will select the `for F` impl
//  during type inference and cause type-checking to fail, as matching this
//  second impl would require a coercion)
impl<'a, F> Pattern<'a> for fn(&char) -> bool { ... }

---

This unfortunate characteristic of these methods was mentioned on the PR that moved them, but I can't find any existing discussion on the possibility of adding impl<'a> Pattern<'a> for fn(&char) -> bool.

[caemor]

Could we maybe change the documenation whenever char::is_aynthing is used to instead use the variant with the closure instead that is working for every (I only tried 2) char::is_anything function?

Benefit would be less confusion when mixing/trying the different char functions.

Disadvantage would be slightly longer (and therefore possible a bit more confusing?) documentation code.