std::path and path-related metabug

[aturon]

In prep for 1.0 stabilization, this metabug collects the various known API issues for std::path. These issues need to be triaged.

Semantic problems:

• path: Windows paths may contain non-utf8-representable sequences path: Windows paths may contain non-utf8-representable sequences #12056
• Windows verbatim paths shouldn't normalize path when .push-ed Windows verbatim paths shouldn't normalize path when .push-ed #11594
• std::path::posix::Path should not automatically normalize std::path::posix::Path should not automatically normalize #14028
• Path::join should concat paths even if the second path is absolute Path::join should concat paths even if the second path is absolute #16507

Ergonomics:

• Path::dirname should return Path instead of [u8] Path::dirname should return Path instead of [u8] #14049
• GenericPath dirname and dirname_str should be removed GenericPath dirname and dirname_str should be removed #10035
• Path::new should return Option vs. failing (if there is a null byte) Path::new should return Option<Path> vs. failing (if there is a null byte) #14048

Implementation:

• pub use and GenericPath versus new sys setup

Missing functionality:

• Make os::make_absolute logic available for arbitrary Paths Make os::make_absolute logic available for arbitrary Paths #5003
• We need fs::realpath We need fs::realpath #11857

[aturon]

cc @kballard -- I thought this would be a good starting point for discussion.

[alexcrichton]

Some other ones off the top of my head:

• There's a lot of *_str() variants, we may want to consolidate these all into on as_str()
• The as_vec method should probably be as_bytes
• The Show trait should probably be implemented on Path
• The BytesContainer trait may need to itself be redesigned, other libstd traits are geared towards one method and this may just want to be geared towards as_bytes()

[aturon]

Update: @kballard and I hammered out a potential new design for the module today, which we hope to file an RFC for in the very near future.

[alexcrichton]

Closing in favor of #20034