chore(deps): update rust crate interprocess to v2

[renovate]

This PR contains the following updates:

Package	Type	Update	Change
interprocess	workspace.dependencies	major	1 -> 2

---

Release Notes

kotauskas/interprocess (interprocess)

v2.1.1

Compare Source

• Removed async Incoming and futures::Stream ("AsyncIterator") implementations on local_socket::traits::Listener implementors – those were actually completely broken, so this change is not breaking in practice and thus does not warrant a bump to 3.0.0
• Fixed ListenerOptionsExt::mode() behavior in umask fallback mode and improved its documentation
• Moved examples to their own dedicated files with the help of the doctest-file crate

v2.1.0: – listeners are now iterators

Compare Source

• Fixes #​49
• Adds Iterator impl on local socket listeners (closes #​64)
• Miscellaneous documentation fixes

v2.0.1: – fix for non-x86 platforms

Compare Source

• Fixed a portability bug that broke the build on Android (#​62, #​63)
• Minor documentation adjustments

v2.0.0

Compare Source

The long-awaited 2.0.0 release brings substantial API enhancements that constitute breaking changes, countless bugfixes, better portability, some future-proofing, a grand refactor and significant scope creep mitigation.

You can upgrade to the new version by changing your Cargo.toml to depend on Interprocess 2.0.0 instead of 1.2.1 and fixing the compilation errors that ensue. The behavioral changes have been made in a manner that avoids the introduction of bugs that successfully compile. The Rustdoc documentation contains all the information that's needed to write code that uses Interprocess 2.0.0, and, by extension, all the information that's needed to port dependents of Interprocess 1.2.1.

• Many bugs have been fixed
  • A comprehensive set of testing utilities has been added to prevent new bugs of the embarrassing variety
• The minimum Rust version has been bumped to 1.75.0, bringing many improvements to both the API and the internals, including RAII of file descriptors and handles and use of RPITITs
• Platform support has been clarified and expanded
• Windows named pipe support has been largely remade:
  • The API has been overhauled to be more idiomatic to Rust with expanded use of generics
  • A "limbo" is now used to prevent in-flight data from being lost when dropping connections
• Small API deficiencies, such as inaccurate types that unnecessarily made invalid options representable, have been corrected
• Windows security descriptors have received a proper API
• Numerous improvements to local sockets
  • They have been made more uniform across platforms – the portability magic that is now in place has eliminated the need to explicitly acknowledge the platform differences in most practical programs
    • Ud-socket local socket listeners now clean up corpse socket files when dropped
  • The new API allows explicit selection of the underlying implementation: every possible backend of local sockets has its own public API, and all of them implement the new local socket traits
  • The name type and the name creation API have been overhauled to facilitate the changes
  • Streams are now splittable by value and readable + writable by reference
• Some functionality has been offloaded to the recvmsg crate – it is now part of the public API of Interprocess
• Ud-sockets have been outsourced to the standard library
• Signal support has been removed in favor of signal_hook

---

Configuration

📅 Schedule: Branch creation - "every weekend" in timezone America/Los_Angeles, Automerge - At any time (no schedule defined).

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR is behind base branch, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR has been generated by Mend Renovate. View repository job log here.

[renovate]

Edited/Blocked Notification

Renovate will not automatically rebase this PR, because it does not recognize the last commit author and assumes somebody else may have edited the PR.

You can manually request rebase by checking the rebase/retry box above.

⚠️ Warning: custom changes will be lost.

[aaronArinder]

do we want unwraps here or something else? Worried about panics (especially without an error message; we should switch to expect if we can't move to something more robust), but also about panics coming from macros making it super, super hard to track down what went wrong

[aaronArinder]

same question about the unwrap here

[dotdat]

I played around with this locally, and it seems like we could get away with something like this in place of a macro:

pub(crate) fn create_socket_name<'s>(raw_socket_name: &'s str) -> std::io::Result<Name<'s>> {
    if GenericFilePath::is_supported() {
        raw_socket_name.clone().to_fs_name::<GenericFilePath>()
    } else {
        raw_socket_name.clone().to_ns_name::<GenericNamespaced>()
    }
}

This would pass the buck on unwrapping or mapping the result to the calling function and solve for aaron's comments below. Passing the buck to the caller for errors is generally something we tend to do instead of panicking.

In the calling code, it's simple enough to do something like:

let socket_name = create_socket_name(&raw_socket_name as &str)?;

This works cleanly because rover maps any error into an anyhow::Error here.

[dotdat]

Is it worth attempting a test for this change to confirm that the behavior is what we expect? We may have to submit the test as a separate PR and then rebase this branch on top of it to confirm, though.

[dotdat]

This appears to work. Ultimately, since we can't query for the filename support anymore, we'll defer to this naming convention on Windows, and the existing naming convention on unix-like systems.