r? @brson

(rust_highfive has picked a reviewer for you, use r? to override)

Discussed during libs triage today the conclusion was that this looks good to go, but could the Default impl be removed?

@alexcrichton It can, although both PathBuf and Box<Path> impl Default now. I'm fine with moving it to another PR though!

Yes let's move to a separate PR

I separated out the Default impl and am working on making the WIP stuff work.

And I've finished the impls! Should work properly; waiting on Travis.

Apparently, Travis is failing because of a type mismatch that doesn't seem to exist? I'm going to test it out locally and see if I can fix it.

So, this seems pretty broken:

error[E0308]: mismatched types
    --> src/libcollections/string.rs:1980:9
     |
1980 |         s.into_string()
     |         ^^^^^^^^^^^^^^^ expected struct `string::String`, found struct `std::string::String`
     |
     = note: expected type `string::String`
                found type `std::string::String`

error[E0308]: mismatched types
    --> src/libcollections/vec.rs:1903:9
     |
1903 |         s.into_vec()
     |         ^^^^^^^^^^^^ expected struct `vec::Vec`, found struct `std::vec::Vec`
     |
     = note: expected type `vec::Vec<T>`
                found type `std::vec::Vec<T>`
     = help: here are some functions which might fulfill your needs:
             - .remove(...)
             - .swap_remove(...)

In what circumstances aren't those the same type? Another note: the implementations for those types do use the types they're asking for, which makes no sense to me.

Got some help on IRC from @thepowersgang (I think); apparently, this is causing the error:

#[cfg(test)]
#[macro_use]
extern crate std;

So right now I disabled the Box -> String and Box -> Vec conversions on tests. Shouldn't be too bad, considering how they just call an already-tested function.

This is ready to merge now, @alexcrichton.

Looks like there may be travis errors?

This seems to have also grown impls since last we looked? The Into impls are a little unfortunate (but seem necessitated by coherence).

I'd be uncomfortable r+'ing all the new impls, so I'll tag with T-libs again.

For Travis, it looks like the errors are from a lack of Clone for the boxed types; I'm going to add those in there too. @.@

I perfectly understand the caution on the new impls; will be waiting for a response from libs. :)

Discussed during triage today, conclusion was that these were good to go.

@bors: r+

📌 Commit 560944b has been approved by alexcrichton

⌛ Testing commit 560944b with merge 71d7b29...

☀️ Test successful - status-appveyor, status-travis
Approved by: alexcrichton
Pushing 71d7b29 to master...