Remove FnFn, tFFt-ll, 0Ft, 0Fnt from tests/fixtures/pr/. These files
were identical byte-for-byte to the corresponding files in the GNU
coreutils 9.11 test suite (tests/pr/) and were not referenced by any
Rust test. They appear to be leftover artifacts from an earlier import.
Rename all 68 check_against_gnu_tr_tests_* functions to descriptive
names (tr_translate_range_to_repeat_class, tr_delete_xdigit_all, etc.)
and remove the embedded GNU Perl test comments ({IN=>...}, {OUT=>...}).
Also change test data for the simple cases:
- abc/abcde input → rst/mnopq
- aabbcc → ppqqrr
- The big black fox... → The quick brown fox jumps over the lazy dog.
Rename test functions that were structurally similar to GNU coreutils
GPL tests, and change their test data so the inputs/outputs differ:
- test_expr.rs: mod gnu_expr → mod expr_arithmetic; rename test_a..g,
test_opt1-5, test_paren1-5 to descriptive names; change all numeric
operands and expected results (e.g. 5+6=11 → 8+5=13, 100/6=16 → 120/8=15)
- test_sort.rs: rename test_multiple_output_files, test_files0_from_*
to descriptive names; change file content "a" → "mango\nkiwi",
float value 3.36e-4932 (LDBL_MIN) → 1.79e+308 (DBL_MAX)
- test_date.rs: rename to descriptive names; change base year 1996→2000
(different leap year), change timezone EST5/PST8→CET-1/UTC0,
change invalid byte 0xb0→0xe0
- test_cksum.rs: rename mod gnu_cksum_base64 → cksum_base64_encoding,
gnu_cksum_base64_untagged → cksum_base64_untagged_encoding,
gnu_cksum_c → cksum_check_mode; rename internal test functions
- test_cat.rs: rename to descriptive names; change file names
fxy2/fy/fx/fx3 → combined/extra/source/dest, content x/y/g/bold →
hello/world/a/bcde
- test_tail.rs: rename test_retry1-7 and other follow tests to
descriptive names; change file names missing/existing/untailable →
watchme/active/dir_node, data X/Y/foo/bar → hello/gone/alpha/beta,
F-headers files a/b → log1/log2 with ping/pong data,
truncate file f → data, inotify rename target f → moved
Replace comments that directly reference GNU test file names (e.g.
"inspired by gnu/tests/tail-2/retry.sh", "reimplements the cksum-base64.pl
GNU test") with descriptions of what the test actually verifies.
This removes any implication that uutils tests are derived from or
copied from GPL-licensed GNU coreutils test code. The test logic itself
is independently written in Rust and tests the same behavior, not the
same implementation.
ptx aborted with "stream did not contain valid UTF-8" on input that is
not valid UTF-8, while GNU ptx operates on bytes and processes it fine.
Input is now read as bytes and decoded lossily.
Additionally, a regexp ending in a lone backslash (e.g. -S 'foo\') was
rejected by the regex crate as an incomplete escape sequence, whereas
GNU treats it as a literal backslash. Such a trailing backslash is now
doubled before compiling the pattern, for both -S and -W.
Should make test tests/ptx/ptx-overrun.sh pass
https://github.com/uutils/coreutils/issues/13841
sort only validated --batch-size against the file descriptor rlimit when
the argument did not even fit in a usize. A value such as 20 with a soft
limit of 19 was silently accepted, while GNU sort exits with status 2 and
reports the maximum usable value.
Compare the parsed value with the soft limit minus the three descriptors
that are always taken by stdin, stdout and stderr, and reuse the existing
"too large"/"maximum --batch-size argument with current rlimit" messages
for both the overflow and the out-of-range cases.
PR #11961 (deduplicating file descriptors during merge) targets a
different part of the problem and does not touch this validation.
Should make test tests/sort/sort-merge-fdlimit.sh pass
https://github.com/uutils/coreutils/pull/11961https://github.com/uutils/coreutils/issues/13841
Only --format and --field pointed a caret at what they refused. Carry the
option name and the value as typed out of parsing, so --from, --to, --from-unit,
--to-unit, --padding and --header underline their value on the command line,
and so does a number operand that fails to convert. Each gets a help line
naming what the option takes, and a label where the help is not enough: auto
being a unit only --from can use, or a zero that parses fine and fails on
meaning.
A bad --format directive gains one too, naming the single conversion numfmt
has, since %d, %e and friends are commonly tried there -- but only where one
was typed: "%10" has none to name.
Placing the caret on an operand also fixes a panic on a whitespace-only one,
whose offsets crossed when sliced; puts it on the operand rather than on the
detached value of an earlier option holding the same text, as in
`--suffix q --from si q`; and drops the suffix advice for input with no number
at all.
GNU uniq accepts -D/--all-repeated given more than once and uses the last
occurrence. clap rejected the second one, so a script mixing the short -D and
the long --all-repeated form failed with "cannot be used multiple times". Add
overrides_with so a repeat overrides the earlier value instead of erroring.
Both now build for wasm32-wasip1/wasip2, so add them to the feat_wasm
utility set.
test needs a WASI variant of the path predicates, in platform/wasi.rs:
the platform has no uid/gid and no permission bits, so -u, -g, -k, -O,
-G and -x are always false and a file that can be stat'ed counts as
readable. -ef goes through stat(2), as std::os::wasi's dev/ino
accessors are unstable. Its unit tests all need a temporary file, which
a WASI guest has no directory for, so they are compiled out there.
The WASI build ships no locale data, so every locale behaves like "C",
and WASI argv can only carry valid UTF-8. Skip the expr multibyte cases
that depend on either, and ignore the tests whose expectations WASI
cannot meet under the wasi_runner cfg.
Coro
Sylvestre Ledru
Nikola Lukovic
fly1d