Simplify output capturing

[m-ou-se]

This is a sequence of incremental improvements to the unstable/internal set_panic and set_print mechanism used by the test crate:

1. Remove the LocalOutput trait and use Arc<Mutex<dyn Write>> instead of Box<dyn LocalOutput>. In practice, all implementations of LocalOutput were just Arc<Mutex<..>>. This simplifies some logic and removes all custom Sink implementations such as library/test/src/helpers/sink.rs. Also removes a layer of indirection, as the outermost Box is now gone. It also means that locking now happens per write_fmt, not per individual write within. (So "{} {}\n" now results in one lock(), not four or more.)

2. Since in all cases the dyn Writes were just Vec<u8>s, replace the type with Arc<Mutex<Vec<u8>>>. This simplifies things more, as error handling and flushing can be removed now. This also removes the hack needed in the default panic handler to make this work with ::realstd, as (unlike Write) Vec<u8> is from alloc, not std.

3. Replace the RefCells by regular Cells. The RefCells were mostly used as mem::replace(&mut *cell.borrow_mut(), something), which is just Cell::replace. This removes an unecessary bookkeeping and makes the code a bit easier to read.

4. Merge set_panic and set_print into a single set_output_capture. Neither the test crate nor rustc (the only users of this feature) have a use for using these separately. Merging them simplifies things even more. This uses a new function name and feature name, to make it clearer this is internal and not supposed to be used by other crates.

Might be easier to review per commit.

[rust-highfive]

r? @KodrAus

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

[KodrAus]

Merge set_panic and set_print into a single set_output_capture

I think these are two unstable APIs that are actually relied upon out in the wild. We have reserved the right to tinker with them, but I'd be interested to see how widely they're actually used! I'll kick off a crater run to see what impact this has.

@craterbot check

[craterbot]

🚨 Error: missing start toolchain

🆘 If you have any trouble with Crater please ping @rust-lang/infra!
ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[KodrAus]

@bors try

[bors]

⌛ Trying commit b85533df13416e98e40f4ce7ac3661b72f24f109 with merge 3e22d1f9053c0fc87629b3c6d5d9567856368e5d...

[m-ou-se]

I think these are two unstable APIs that are actually relied upon out in the wild.

Those APIs were already broken recently by #75172 / #78227. Those didn't get a crater run, though. Searching for set_print and set_panic on GitHub mostly gives results inside forks of Rust itself.

(These functions aren't just unstable, they are doc(hidden) too.)

[bors]

☀️ Try build successful - checks-actions
Build commit: 3e22d1f9053c0fc87629b3c6d5d9567856368e5d (3e22d1f9053c0fc87629b3c6d5d9567856368e5d)

[m-ou-se]

Starting crater with start=beta instead of the parent commit of this PR, because the function signatures were already changed/broken two weeks ago. (Hopefully that won't report too many unrelated changes.)

@craterbot check start=beta

[craterbot]

🚨 Error: missing end toolchain

🆘 If you have any trouble with Crater please ping @rust-lang/infra!
ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[m-ou-se]

@craterbot check start=beta end=try#3e22d1f9053c0fc87629b3c6d5d9567856368e5d

[craterbot]

👌 Experiment pr-78714 created and queued.
🔍 You can check out the queue and this experiment's details.

ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[craterbot]

🚧 Experiment pr-78714 is now running

ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[craterbot]

🎉 Experiment pr-78714 is completed!
📊 379 regressed and 4349 fixed (129870 total)
📰 Open the full report.

⚠️ If you notice any spurious failure please add them to the blacklist!
ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[m-ou-se]

That crater run resulted in quite a few irrelevant failures, because it compaired against beta. These are the only four that had errors in their logs about missing set_print or set_panic:

• wasm_glue:
  This crate's main purpose is to use set_print and set_panic to work around some issues with stdout/stderr on wasm. This crate doesn't seem to be very popular though.
• cis198-2016s/demo-xcompile-android:
  Fails because it depends on an old version of the deprecated android-rs-glue project that used set_print/set_panic. Newer versions and the project that superseded it don't use those functions.
• reitermarkus/lndir and rvolz/rustrt:
  These both fail because of their dependency on an old version of cucumber_rust. Versions since may 2019 don't use set_print or set_panic anymore.

[Mark-Simulacrum]

FWIW we have access to by-commit artifacts so you can get a smaller commit range, but ultimately it probably wouldn't help too much. If we want a better run I would probably revert the breakage in a pr, try commit that, then compare against that in crater.

Let me know if you need more help here.

[m-ou-se]

In this case I knew what the error would be, so it was as simple as just downloading the build-fail tar.gz from crater and grepping for set_panic and set_print. Only four crates matched.

[KodrAus]

That's a lot fewer than I was expecting! Thanks for looking into this @m-ou-se

[KodrAus]

Ok, this looks good to me. I was worried about churning an unstable hack that I thought was pretty widely relied on, but it turns out I was wrong!

@bors r+

[bors]

📌 Commit aff7bd6 has been approved by KodrAus