Tracking Issue for dir_entry_ext2

[arennow]

Feature gate: #![feature(dir_entry_ext2)]

This is a tracking issue for a new method on std::fs::DirEntry on UNIX platforms via a new sealed variant of the std::os::unix::fs::DirEntryExt trait.

Public API

#[unstable(feature = "dir_entry_ext2")]
pub trait DirEntryExt2: Sealed {
    fn file_name_ref(&self) -> &OsStr;
}

Steps / History

• Implementation: Add std::os::unix::fs::DirEntryExt2::file_name_ref(&self) -> &OsStr #83581
• Final commenting period (FCP)
• Stabilization PR

Unresolved Questions

• What is an appropriate name for this new trait?

[marcospb19]

Hey, @arennow and @m-ou-se, is it possible to have an owned version of this function?

(self) -> OsString

Something like this in the trait

#[unstable(feature = "dir_entry_ext2")]
pub trait DirEntryExt2: Sealed {
    fn file_name_ref(&self) -> &OsStr;
    fn into_file_name(self) -> OsStr;
}

[tavianator]

@marcospb19 An owned version is already available from DirEntry::file_name()

[mic006]

Hi @arennow and @m-ou-se

Is it planned to make this change stable in a coming release ?
The change seems to be a good optimization (typically for sorting DirEntry-s), without any risk of issue.

[cgwalters]

Another year later: I also came across this one after writing some code to sort directory entries and thinking about the allocations involved.

It looks like we got hung up previously on the extension trait sealing: #83581 (comment)

I guess we need another crater run?