base32, base64, basenc: drop the "error: " from the invalid-input message

[Socialpranker]

The invalid-input diagnostic carried a literal error: on top of the
program name that uucore already prints:

$ printf 'not base64!' | base64 -d            # GNU
base64: invalid input
$ printf 'not base64!' | base64 -d            # uutils, before
base64: error: invalid input

$ printf 'abc' | basenc --z85                 # GNU
basenc: invalid input (length must be multiple of 4 characters)
$ printf 'abc' | basenc --z85                 # uutils, before
basenc: error: invalid input (length must be multiple of 4 characters)

No other diagnostic in these utilities spells the word "error" — the
neighbouring messages are basenc: missing encoding type,
base64: extra operand, and so on — and none of the GNU ones do either.

The fix

Dropping the prefix from the 13 message literals in
src/uucore/src/lib/features/encoding.rs and
src/uu/base32/src/base_common.rs, and from the seven test assertions
that pinned the old wording. Nothing else changes: same error kind, same
exit status 1, same trailing detail in the z85 length message.

Not addressed here: GNU also writes the successfully decoded prefix to
stdout before failing, which uutils does not — that is #6008 and is left
alone.

How the GNU behavior was established

By running the installed GNU coreutils 9.11 binaries (Homebrew, gbase32,
gbase64, gbasenc) as a black box on undecodable input across
--base64, --base64url, --base32, --base32hex, --base16, --z85
and the plain base32 -d / base64 -d forms, and diffing stderr against
uutils. I did not read GNU coreutils source.

Testing

• The seven existing assertions in tests/by-util/test_base32.rs,
  test_base64.rs and test_basenc.rs now pin the GNU wording, so they
  are the mutation check: reverting the two source files alone makes all
  seven fail.
• cargo test --features "base32 base64 basenc" --test tests test_base:
  119 passed, 0 failed.
• cargo clippy -p uu_base32 -p uu_base64 -p uu_basenc --all-targets -- -D warnings:
  clean.
• cargo fmt --check: clean.
• Differential A/B against GNU coreutils 9.11 over 713 invocations across
  28 utilities: mismatches 79 -> 74 (base32 1 -> 0, basenc 6 -> 2), no
  other bucket moved. The four that remain in these utilities are the
  partial-output difference of basenc: emit partial output on invalid input #6008.

Disclosure

Prepared with AI assistance (Claude Opus 5, via Claude Code), per the AI
policy in CONTRIBUTING.md. Every GNU behavior quoted above came from
running the installed binaries, not from reading GPL source. All testing
was run locally.

[Socialpranker]

Pushed one more commit: util/build-gnu.sh patched tests/basenc/basenc.pl to add error: to the upstream expectation, so that sed is now obsolete and removed — the GNU test expects what uutils prints.

For the record, the one red job here (Build (ubuntu-latest, i686-unknown-linux-gnu, …)) is test_numfmt::test_unit_hyphen_leading_as_separate_arg panicking on failed to write to stdin of child: Broken pipe — a test-harness race in an unrelated utility, not something this change can reach.

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/cut/bounded-memory (fails in this run but passes in the 'main' branch)
Skip an intermittent issue tests/tail/retry (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/tail/tail-n0f (passes in this run but fails in the 'main' branch)
Note: The gnu test tests/csplit/csplit-heap is now being skipped but was previously passing.

[anastygnome]

Could you please squash your commits ? Thanks

[anastygnome]

This should be localised as done in other utils

[anastygnome]

SHould be localised

[Socialpranker]

Localised as base-common-invalid-input in base_common.rs/.ftl, sha a8d7da6.

[anastygnome]

should be localised too

[Socialpranker]

Localised as encoding-error-invalid-input/encoding-error-invalid-input-z85-length in uucore/encoding.rs/.ftl, sha a8d7da6.

[Socialpranker]

Squashed to a single commit, sha a8d7da6.

[codspeed-hq]

Merging this PR will regress 2 benchmarks

⚡ 6 improved benchmarks
❌ 2 regressed benchmarks
✅ 359 untouched benchmarks
⏩ 50 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	mv_directory	8.2 ms	8.5 ms	-3.52%
❌	Simulation	wc_bytes_synthetic[500]	161.7 µs	167.6 µs	-3.52%
⚡	Simulation	mv_force_overwrite	177.9 ms	147.5 ms	+20.61%
⚡	Simulation	hostname_ip_lookup[100000]	179 µs	164.4 µs	+8.85%
⚡	Simulation	cksum_crc	34 ms	32.5 ms	+4.69%
⚡	Simulation	cksum_default	33.7 ms	32.4 ms	+3.83%
⚡	Simulation	cksum_raw_output	33.7 ms	32.7 ms	+3.14%
⚡	Simulation	cksum_multiple_files	50.4 ms	48.9 ms	+3.05%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing Socialpranker:base-drop-error-prefix (a8d7da6) with main (9ff4114)}

Footnotes

1. 50 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[anastygnome]

Flaky CI