Add IsTerminal trait to determine if a descriptor or handle is a terminal

[joshtriplett]

The UNIX implementation uses isatty. The Windows implementation uses
the same logic the atty crate uses, including the hack needed to
detect msys terminals.

Implement this trait for Stdin/Stdout/Stderr/File on all
platforms. On Unix, implement it for BorrowedFd/OwnedFd. On Windows,
implement it for BorrowedHandle/OwnedHandle.

Based on #91121

Co-authored-by: Matt Wilkinson [email protected]

[rust-highfive]

Hey! It looks like you've submitted a new PR for the library teams!

If this PR contains changes to any rust-lang/rust public library APIs then please comment with @rustbot label +T-libs-api -T-libs to tag it appropriately. If this PR contains changes to any unstable APIs please edit the PR description to add a link to the relevant API Change Proposal or create one if you haven't already. If you're unsure where your change falls no worries, just leave it as is and the reviewer will take a look and make a decision to forward on if necessary.

Examples of T-libs-api changes:

• Stabilizing library features
• Introducing insta-stable changes such as new implementations of existing stable traits on existing stable types
• Introducing new or changing existing unstable library APIs (excluding permanently unstable features / features without a tracking issue)
• Changing public documentation in ways that create new stability guarantees
• Changing observable runtime behavior of library APIs

[rust-highfive]

r? @kennytm

(rust-highfive has picked a reviewer for you, use r? to override)

[the8472]

If we had the view-from-borrowed thing then we don't need the sealing complications and instead one would do file.as_view::<IsTerminal>().is_terminal() where IsTerminal is an fd-wrapping struct like File.

[joshtriplett]

sealing complications

I just finished implementing sealing after all, which turned out to be possible.

[the8472]

This could also be a portability hazard. On unix you can now do tcp_socket.is_terminal() but it would fail to compile on windows because sockets and handles are separate.

I think it would be better if we added File::is_terminal (since any file could be a terminal anyway. /dev/tty, conin$) and then people would have to acknowledge the platform-dependence themselves by doing.

let is_terminal = false;
#[cfg(windows)]
let is_terminal = stdin.as_view::<File>().is_terminal()?;
#[cfg(unix)]
let is_terminal = stdin.as_view::<File>().is_terminal()?;

This isn't too terrible since one doesn't need cfg_if, only shadowing.

[joshtriplett]

@the8472 I continue to be quite skeptical of the "File view" mechanism, as opposed to BorrowedFd and OwnedFd and similar.

However, we could change this to implement IsTerminal for Stdin/Stdout/Stderr/File on every platform, and then implement it for BorrowedFd and OwnedFd on Unix, and BorrowedHandle and OwnedHandle on Windows. That would then mean you can unconditionally and portably call it on Stdin/Stdout/Stderr/File, while if you want to call it on an AsFd you have to call .as_fd().is_terminal() and import AsFd with associated target-specific code.

[the8472]

I continue to be quite skeptical of the "File view" mechanism, as opposed to BorrowedFd and OwnedFd and similar.

I have shifted my position a bit, wrappers should have to explicitly opt-in to this kind of conversion but I see File as a prime candidate for that because it does really basic operations that should either be valid or harmlessly error on any file descriptor (read, write, stat, fsync, seek, chmod). At least on unix is_terminal is just another flavor of stat. pipes, sockets and ttys can show up in the filesystem anyway.

I think the conversion is better because we have those structs and useful methods implemented on them anyway. Similarly something might pass a socket as std-in and if you want to do socket-ops on it you need to borrow it as a TcpStream for example.

But maybe you're right and this is a mistake of the past and the basic operations should have been implemented on BorrowedFd and only stateful operations such as iterating a directory via getdents would live on wrapper types

However, we could change this to implement IsTerminal for Stdin/Stdout/Stderr/File on every platform, and then implement it for BorrowedFd and OwnedFd on Unix, and BorrowedHandle and OwnedHandle on Windows.

Not how I'd do it, but that works too.

[matthiaskrgr]

@bors r- rollup=iffy failed in a rollup
#102814 (comment)

[joshtriplett]

Rebased on top of #102847 .

@bors r=thomcc

[bors]

📌 Commit 97d438c has been approved by thomcc

It is now in the queue for this repository.

[bors]

⌛ Testing commit 97d438c with merge 8154955...

[bors]

☀️ Test successful - checks-actions
Approved by: thomcc
Pushing 8154955 to master...

[rust-timer]

Finished benchmarking commit (8154955): comparison URL.

Overall result: no relevant changes - no action needed

@rustbot label: -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean1	range	count2
Regressions ❌ (primary)	2.4%	[1.5%, 3.4%]	2
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-3.2%	[-3.9%, -2.6%]	3
Improvements ✅ (secondary)	-3.1%	[-5.3%, -2.1%]	29
All ❌✅ (primary)	-1.0%	[-3.9%, 3.4%]	5

Cycles

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean1	range	count2
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.5%	[-2.8%, -2.1%]	2
Improvements ✅ (secondary)	-4.0%	[-5.9%, -2.1%]	2
All ❌✅ (primary)	-2.5%	[-2.8%, -2.1%]	2

Footnotes

1. the arithmetic mean of the percent change ↩ ↩²

2. number of relevant changes ↩ ↩²