feat: add /dev/null handling

[puffnfresh]

This allows some scripts to run on Windows, which otherwise fail because the /dev/null device does not exist. This is similar to the existing special /dev/fd handling which Brush already contains.

[oech3]

I think it is something Cygwin should care about instead of shell.

[puffnfresh]

@oech3 is that an argument specifically for /dev/null handling? Brush already does /dev/fd and Bash provides a few more. Where should we draw the line?

I think /dev/null is such a common idiom, it's a shame to have to use a Cygwin environment for many scripts which otherwise run fine on Windows.

[oech3]

Hmm... I did't know that bash has such hacks.

[oech3]

Are you planning to suggest this to upstram bash too?

[reubeno]

@puffnfresh thanks for the contribution! It's good to know that some simple changes would unblock more scripts from running natively on Windows under brush.

I also support the idea of being able to use more scripts in a cross-platform way without requiring emulation layers like Cygwin.

Regarding the change, you're right that we have precedent with /dev/fd handling. It wasn't ideal but we were forced to do that because of brush's design of abstracting and decoupling the file descriptors addressable within the shell from the host process's native file descriptors even on platforms like Linux and macOS.

I'm supportive of special handling for /dev/null and similar special files that are ubiquitously accessed in scripts, but we should scope that emulation to the platforms that require it (e.g., Windows).

We use the sys module to abstract platform-specific behaviors. What do you think of adding a new call into the sys module in this code path to give platform-specific code an opportunity to optionally handle special files. We would have the Unix-y platforms implement a no-op, but Windows and others could have a special case there for /dev/null.

This would allow Unix platforms to open the real null file, which is likely to look more consistent with other shells, while still enabling scripts to achieve the desired effect elsewhere.

[github-actions]

Public API changes for crate: brush-core

Added items

+pub const fn brush_core::sys::fs::try_open_special_file(_path: &std::path::Path) -> core::option::Option<core::result::Result<std::fs::File, std::io::error::Error>>

Performance Benchmark Report

Benchmark name	Baseline (μs)	Test/PR (μs)	Delta (μs)	Delta %
clone_shell_object	17.28 μs	17.45 μs	0.16 μs	🟠 +0.95%
eval_arithmetic	0.15 μs	0.15 μs	-0.01 μs	⚪ Unchanged
expand_one_string	1.67 μs	1.50 μs	-0.17 μs	⚪ Unchanged
for_loop	27.63 μs	27.70 μs	0.07 μs	⚪ Unchanged
full_peg_complex	57.93 μs	58.15 μs	0.23 μs	⚪ Unchanged
full_peg_for_loop	6.19 μs	6.21 μs	0.02 μs	⚪ Unchanged
full_peg_nested_expansions	16.75 μs	16.82 μs	0.08 μs	⚪ Unchanged
full_peg_pipeline	4.26 μs	4.31 μs	0.04 μs	🟠 +1.06%
full_peg_simple	1.83 μs	1.85 μs	0.03 μs	🟠 +1.48%
function_call	2.96 μs	2.96 μs	0.01 μs	⚪ Unchanged
instantiate_shell	55.82 μs	54.21 μs	-1.61 μs	🟢 -2.89%
instantiate_shell_with_init_scripts	24740.62 μs	25149.68 μs	409.06 μs	🟠 +1.65%
parse_peg_bash_completion	2071.74 μs	2077.14 μs	5.40 μs	⚪ Unchanged
parse_peg_complex	19.76 μs	19.98 μs	0.22 μs	🟠 +1.11%
parse_peg_for_loop	2.00 μs	1.99 μs	-0.01 μs	⚪ Unchanged
parse_peg_pipeline	2.12 μs	2.17 μs	0.04 μs	🟠 +2.12%
parse_peg_simple	1.12 μs	1.13 μs	0.00 μs	⚪ Unchanged
run_echo_builtin_command	16.66 μs	16.47 μs	-0.19 μs	⚪ Unchanged
tokenize_sample_script	3.40 μs	3.38 μs	-0.03 μs	⚪ Unchanged

Code Coverage Report: Only Changed Files listed

Package	Base Coverage	New Coverage	Difference
brush-core/src/shell/fs.rs	🟢 98.21%	🟢 97.39%	🔴 -0.82%
brush-core/src/sys/unix/fs.rs	🟠 61.32%	🟠 62.39%	🟢 1.07%
brush-test-harness/src/comparison.rs	🟢 80.77%	🟢 76.92%	🔴 -3.85%
brush-test-harness/src/execution.rs	🟢 84.04%	🟢 80.85%	🔴 -3.19%
brush-test-harness/src/reporting.rs	🔴 43.66%	🔴 11.62%	🔴 -32.04%
brush-test-harness/src/runner.rs	🟢 82.81%	🟢 82.21%	🔴 -0.6%
brush-test-harness/src/util.rs	🟢 92.31%	🟠 61.54%	🔴 -30.77%
Overall Coverage	🟢 74.89%	🟢 74.36%	🔴 -0.53%

Minimum allowed coverage is 70%, this run produced 74.36%

Test Summary: bash-completion test suite

Outcome	Count	Percentage
✅ Pass	1583	75.06
❗️ Error	18	0.85
❌ Fail	154	7.30
⏩ Skip	339	16.07
❎ Expected Fail	13	0.62
✔️ Unexpected Pass	2	0.09
📊 Total	2109	100.00

[reubeno]

Change look good to me, @puffnfresh. Thanks for the update!

Looks like the PR checks flagged formatting issues -- and you'll need to rebase/merge latest changes.

[reubeno]

Merged! Thanks for working on this, @puffnfresh. There was just one last clippy warning that I fixed for you so I could get it merged.

If you end up using this on Windows, would be great to hear what your experience is like / what you're able to do. If you're up for it, feel free to post issues / discussions here on GitHub or on our Discord.