replace_*_zero_width tests fail with Err(Error(0)) (panic on unwrap)

[Boria138]

After a successful build, the tests fail; here is part of the spec

Requires: libicu%icu_ver

%description
An editor that pays homage to the classic MS-DOS Editor, but with a modern interface and input controls similar to VS Code.

%prep
%setup -a1
%patch -p1

cat >.cargo/config <<EOF
[source.crates-io]
replace-with = "vendored-sources"
[source.vendored-sources]
directory = "vendor"
EOF

%build
export EDIT_CFG_ICUUC_SONAME=libicuuc.so.%icu_ver
export EDIT_CFG_ICUI18N_SONAME=libicui18n.so.%icu_ver
export EDIT_CFG_ICU_RENAMING_VERSION=%icu_ver

%rust_build

%install
install -Dm 755 target/release/edit -t %buildroot%_bindir

%check
%rust_test

%files
%_bindir/edit
%doc LICENSE

For icu_ver 7.6, Rust version 1.95 and 1.2.0 build without issues, but 2.0.0 does not. The problem might be that I removed RUSTC_BOOTSTRAP=1, and the tests require nightly features. Here is the test error itself

Finished release profile [optimized + debuginfo] target(s) in 16.90s
[00:00:27]      Running unittests src/lib.rs (target/release/deps/edit-537b5e8384b30508)
[00:00:27] 
[00:00:27] running 43 tests
[00:00:27] test icu::tests::init ... ignored
[00:00:27] test base64::tests::test_basic ... ok
[00:00:27] test buffer::navigation::test::test_word_navigation ... ok
[00:00:27] test icu::tests::test_compare_strings_ascii ... ok
[00:00:27] test json::tests::test_bool ... ok
[00:00:27] test buffer::tests::replace_all_zero_width ... FAILED
[00:00:27] test buffer::tests::replace_one_zero_width ... FAILED
[00:00:27] test json::tests::test_control_chars ... ok
[00:00:27] test json::tests::test_max_depth ... ok
[00:00:27] test json::tests::test_array ... ok
[00:00:27] test json::tests::test_comments ... ok
[00:00:27] test json::tests::test_nested ... ok
[00:00:27] test json::tests::test_null ... ok
[00:00:27] test json::tests::test_number ... ok
[00:00:27] test json::tests::test_invalid_json ... ok
[00:00:27] test json::tests::test_object ... ok
[00:00:27] test json::tests::test_trailing_comma ... ok
[00:00:27] test json::tests::test_string ... ok
[00:00:27] test json::tests::test_unicode ... ok
[00:00:27] test oklab::tests::test_blending ... ok
[00:00:27] test path::tests::test_unix ... ok
[00:00:27] test simd::lines_bwd::test::seeks_to_start ... ok
[00:00:27] test simd::memchr2::tests::test_basic ... ok
[00:00:27] test simd::memchr2::tests::test_empty ... ok
[00:00:27] test simd::memchr2::tests::test_with_offset ... ok
[00:00:27] test unicode::measurement::test::test_any_wrap ... ok
[00:00:27] test simd::memchr2::tests::test_page_boundary ... ok
[00:00:27] test unicode::measurement::test::test_any_wrap_wide ... ok
[00:00:27] test unicode::measurement::test::test_crlf ... ok
[00:00:27] test simd::lines_fwd::test::pseudo_fuzz ... ok
[00:00:27] test unicode::measurement::test::test_exact_wrap ... ok
[00:00:27] test unicode::measurement::test::test_force_wrap ... ok
[00:00:27] test unicode::measurement::test::test_force_wrap_column ... ok
[00:00:27] test unicode::measurement::test::test_force_wrap_wide ... ok
[00:00:27] test unicode::measurement::test::test_measure_forward_chunk_boundaries ... ok
[00:00:27] test unicode::measurement::test::test_measure_forward_clipped_wide_char ... ok
[00:00:27] test unicode::measurement::test::test_measure_forward_newline_start ... ok
[00:00:27] test unicode::measurement::test::test_measure_forward_tabs ... ok
[00:00:27] test unicode::measurement::test::test_measure_forward_word_wrap ... ok
[00:00:27] test unicode::measurement::test::test_strip_newline ... ok
[00:00:27] test unicode::measurement::test::test_wrap_tab ... ok
[00:00:27] test unicode::measurement::test::test_wrapped_cursor_can_seek_backward ... ok
[00:00:27] test simd::lines_bwd::test::pseudo_fuzz ... ok
[00:00:27] 
[00:00:27] failures:
[00:00:27] 
[00:00:27] ---- buffer::tests::replace_all_zero_width stdout ----
[00:00:27] 
[00:00:27] thread 'buffer::tests::replace_all_zero_width' (88153) panicked at crates/edit/src/buffer/mod.rs:3178:10:
[00:00:27] called Result::unwrap() on an Err value: Error(0)
[00:00:27] 
[00:00:27] ---- buffer::tests::replace_one_zero_width stdout ----
[00:00:27] 
[00:00:27] thread 'buffer::tests::replace_one_zero_width' (88154) panicked at crates/edit/src/buffer/mod.rs:3160:14:
[00:00:27] called Result::unwrap() on an Err value: Error(0)
[00:00:27] note: run with RUST_BACKTRACE=1 environment variable to display a backtrace
[00:00:27] 
[00:00:27] 
[00:00:27] failures:
[00:00:27]     buffer::tests::replace_all_zero_width
[00:00:27]     buffer::tests::replace_one_zero_width
[00:00:27] 
[00:00:27] test result: FAILED. 40 passed; 2 failed; 1 ignored; 0 measured; 0 filtered out; finished in 0.00s
[00:00:27] 
[00:00:27] error: test failed, to rerun pass --lib
[00:00:27]      Running unittests src/bin/edit/main.rs (target/release/deps/edit-17c0b67aa99d6b88)
[00:00:27] 
[00:00:27] running 1 test
[00:00:27] test documents::tests::test_parse_last_numbers ... ok
[00:00:27] 
[00:00:27] test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
[00:00:27] 
[00:00:27]    Doc-tests edit
[00:00:28] 
[00:00:28] running 2 tests
[00:00:28] test crates/edit/src/tui.rs - tui::Context<'a,'_>::styled_label_begin (line 1963) ... ok
[00:00:28] test crates/edit/src/tui.rs - tui (line 91) ... ok
[00:00:28] 
[00:00:28] test result: ok. 2 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
[00:00:28] 
[00:00:28] all doctests ran in 1.66s; merged doctests compilation took 1.66s
[00:00:28] error: 1 target failed:
[00:00:28]     --lib
[00:00:28] error: Bad exit status from /usr/src/tmp/rpm-tmp.24248 (%check)
[00:00:28] 
[00:00:28] 
[00:00:28] RPM build errors:
[00:00:28]     Bad exit status from /usr/src/tmp/rpm-tmp.24248 (%check)
[00:00:28] Command exited with non-zero status

[Boria138]

Oh, looks like this is a duplicate

#775

[Boria138]

I took a closer look at d14e24b version 2.0.0, but the tests are failing

[lhecker]

Err(Error(0)) is my way of piggy-backing an "ICU failed loading" status code (aka ICU_MISSING_ERROR).

This means your EDIT_CFG_ICUUC_SONAME, etc., configuration appears to have failed.

[Boria138]

Err(Error(0)) is my way of piggy-backing an "ICU failed loading" status code (aka ICU_MISSING_ERROR).

This means your EDIT_CFG_ICUUC_SONAME, etc., configuration appears to have failed.

It looks like an error in the tests; here's what happens during the build

[00:00:03] + export EDIT_CFG_ICUUC_SONAME=libicuuc.so.76
[00:00:03] + EDIT_CFG_ICUUC_SONAME=libicuuc.so.76
[00:00:03] + export EDIT_CFG_ICUI18N_SONAME=libicui18n.so.76
[00:00:03] + EDIT_CFG_ICUI18N_SONAME=libicui18n.so.76
[00:00:03] + export EDIT_CFG_ICU_RENAMING_VERSION=76
[00:00:03] + EDIT_CFG_ICU_RENAMING_VERSION=76

Here are the package installation logs from the build system

<13>Apr 29 15:34:35 rpmi: libicu76-1:7.6.1 installed

Here is the list of files in the package

$ rpm -ql libicu76
/usr/lib64/libicudata.so.76
/usr/lib64/libicudata.so.76.1
/usr/lib64/libicui18n.so.76
/usr/lib64/libicui18n.so.76.1
/usr/lib64/libicuio.so.76
/usr/lib64/libicuio.so.76.1
/usr/lib64/libicutest.so.76
/usr/lib64/libicutest.so.76.1
/usr/lib64/libicutu.so.76
/usr/lib64/libicutu.so.76.1
/usr/lib64/libicuuc.so.76
/usr/lib64/libicuuc.so.76.1
/usr/share/doc/libicu76-7.6.1
/usr/share/doc/libicu76-7.6.1/LICENSE
/usr/share/doc/libicu76-7.6.1/readme.html

Until release 2.0.0, I was building from your PR #495 and everything worked

[DHowett]

@lhecker I think this is the first test that actually requires icu... perhaps that is a lead

[DHowett]

So @Boria138, if you take the edit binary generated by your build and ignore the failing tests, does the Find feature work? Did it work before d14e24b?

I'm thinking that the test failure is a canary in your coal mine. ;P

[Boria138]

So @Boria138, if you take the edit binary generated by your build and ignore the failing tests, does the Find feature work? Did it work before d14e24b?

I'm thinking that the test failure is a canary in your coal mine. ;P

The search function worked in version 1.2.0 with PR 495 applied as a patch; I haven't checked yet whether it works now if I ignore the tests.

[lhecker]

If you tell us more about the distribution, environment and build scripts you're using, we could try to reproduce the issue ourselves. Otherwise, it will be difficult to diagnose this issue.

[Boria138]

if you take the edit binary generated by your build and ignore the failing tests, does the Find feature work?

I just checked, and everything is working—so the problem must be somewhere in the tests, not in the build itself.

[Boria138]

If you tell us more about the distribution, environment and build scripts you're using, we could try to reproduce the issue ourselves. Otherwise, it will be difficult to diagnose this issue.

cat /etc/os-release
NAME="ALT"
VERSION="20260316"
ID=altlinux
VERSION_ID=20260316
PRETTY_NAME="ALT Regular"
VARIANT="Regular"
ANSI_COLOR="1;33"
CPE_NAME="cpe:/o:alt:sisyphus:20260316"
BUILD_ID="Sisyphus 20240122"
ALT_BRANCH_ID="sisyphus"
HOME_URL="http://en.altlinux.org"
BUG_REPORT_URL="https://bugs.altlinux.org/"
LOGO=altlinux

[Boria138]

Did it work before d14e24b?

The tests pass when this commit is reverted, which means the problem lies in it

[DHowett]

It looks like the check phase in rpmbuild doesn't use any of the environment variables set in the build phase, and the entire project gets rebuilt during check.

Therefore, the build gets the EDIT_CFG_ICU* environment variables and the tests do not. d14e24b "regressed" this because it adds the first test-time dependency on ICU.

Add the same three environment variables at the top of your %check block.

The build will succeed after that:

test result: ok. 42 passed; 0 failed; 1 ignored; 0 measured; 0 filtered out; finished in 0.05s

...

Wrote: /home/dustin/RPM/SRPMS/msedit-1-1.src.rpm (w2.lzdio)
Wrote: /home/dustin/RPM/RPMS/x86_64/msedit-1-1.x86_64.rpm (w2.lzdio)
[dustin@cpu-3ceba6 edit]$ head -n2 /etc/os-release
NAME="ALT"
VERSION="20260316"

[DHowett]

(and please excuse the bad version number - I had to learn how to write an RPM spec somehow ;P)