refactor: remove the encapsulation for rspack_fs::Error

[CPunisher]

Summary

rspack_fs::Error contains only std::io::Error for two years. So here considering remove the encapsulation.

There could be some thing wrong with ci linux tools, which produce ~1MB larger artifact than the that in my pc with cross-compilation.

Checklist

• Tests updated (or not required).
• Documentation updated (or not required).

[netlify]

✅ Deploy Preview for rspack canceled.

Name	Link
🔨 Latest commit	ec2e044
🔍 Latest deploy log	https://app.netlify.com/projects/rspack/deploys/686ccf6adbfb9b00085a80f2

[github-actions]

📦 Binary Size-limit

Comparing 7e4147c to chore: release v1.4.5 (#10944) by Gengkun

❌ Size increased by 356.80MB from 59.52MB to 416.32MB (⬆️599.48%)

[codspeed-hq]

CodSpeed Performance Report

Merging #10942 will not alter performance

_{Comparing 07-07-refactor/remove-rspack-fs-error (ec2e044) with main (a306b03)}

Summary

✅ 16 untouched benchmarks

[Copilot]

Pull Request Overview

This PR removes the custom rspack_fs::Error wrapper and uses std::io::Error directly, simplifying error handling and aiming to reduce binary size. It updates all filesystem operations to propagate standard I/O errors, replaces to_fs_result conversions with ?, and integrates diagnostic conversions where appropriate.

• Exposed std::io::Error as the filesystem error type and removed the custom Error enum.
• Updated all to_fs_result calls to direct ? propagation and added into_diagnostic() for richer error reports.
• Refactored filesystem implementations (native_fs, memory_fs, intermediate, core utilities) to align with the new error model.

Reviewed Changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
packages/rspack-test-tools/package.json	Added a test-wasm script for WASM-based tests
crates/rspack_storage/src/fs/error.rs	Switched FSError::inner to use std::io::Error directly
crates/rspack_plugin_sri/src/html.rs	Added .into_diagnostic() on read_file
crates/rspack_plugin_dll/src/lib_manifest_plugin.rs	Wrapped async filesystem calls with .into_diagnostic()
crates/rspack_fs/src/native_fs.rs	Removed to_fs_result wrappers; propagate std::io::Error
crates/rspack_fs/src/memory_fs.rs	Replaced custom error construction with Error::new
crates/rspack_fs/src/lib.rs	Removed export of custom error types; aliased to std::io::Error
crates/rspack_fs/src/intermediate.rs	Simplified UTF-8 error return to std::io::Error::other
crates/rspack_fs/src/file_metadata.rs	Simplified metadata calls (modified()?, accessed()?)
crates/rspack_fs/src/error.rs	Replaced the Error enum with a type alias to std::io::Error
crates/rspack_core/src/utils/fs_trim.rs	Changed try_match to return FsResult<bool>
crates/rspack_core/src/resolver/boxfs.rs	Simplified ResolverFileSystem error mappings
crates/rspack_core/src/compiler/mod.rs	Added .into_diagnostic() around various await calls

Comments suppressed due to low confidence (3)

crates/rspack_fs/src/memory_fs.rs:243

• [nitpick] The error message is ungrammatical. Consider using "file does not exist" for clarity and proper tense.

      _ => Err(Error::new(ErrorKind::NotFound, "file not exist")),

crates/rspack_fs/src/memory_fs.rs:255

• [nitpick] This error message could be improved to "file does not exist" to follow standard English usage.

      None => Err(Error::new(ErrorKind::NotFound, "file not exist")),

crates/rspack_fs/src/intermediate.rs:25

• [nitpick] Consider capitalizing "UTF-8" and refining the message to "invalid UTF-8 line" for clarity.

      Err(_) => Err(std::io::Error::other("invalid utf8 line")),

[Copilot]

[nitpick] There are two identical impl WritableFileSystem for NativeFileSystem blocks in this file. Consider merging them into one implementation to reduce duplication.

[CPunisher]

There could be some thing wrong with ci linux tools, which produce ~1MB larger artifact than the that in my pc with cross-compilation. But I have no time to investigate currently.

UPDATE: it seems the ci use GLIBC with newer version