orium/rpds

sparse_array_usize test failures on i686

Closed this issue · 2 comments

mattiaverga commented

Hello,
while packaging rpds in Fedora we encountered some tests failures which happen only on i686 architecture:

...
test map::hash_trie_map::sparse_array_usize::test::test_first ... FAILED
...
test map::hash_trie_map::sparse_array_usize::test::test_pop ... FAILED
...
failures:
---- map::hash_trie_map::sparse_array_usize::test::test_first stdout ----
thread 'map::hash_trie_map::sparse_array_usize::test::test_first' panicked at 'assertion failed: `(left == right)`
Diff < left / right > :
 Some(
<    '�[1;48;5;52;31mb',
>    '�[1;48;5;22;32ma',
 )
', src/map/hash_trie_map/sparse_array_usize/test.rs:85:5
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
---- map::hash_trie_map::sparse_array_usize::test::test_pop stdout ----
thread 'map::hash_trie_map::sparse_array_usize::test::test_pop' panicked at 'assertion failed: `(left == right)`
Diff < left / right > :
 Some(
<    '�[1;48;5;52;31ma',
>    '�[1;48;5;22;32mb',
 )
', src/map/hash_trie_map/sparse_array_usize/test.rs:101:5
failures:
    map::hash_trie_map::sparse_array_usize::test::test_first
    map::hash_trie_map::sparse_array_usize::test::test_pop
test result: FAILED. 257 passed; 2 failed; 0 ignored; 0 measured; 0 filtered out; finished in 3.71s

I am wondering if this is just a rounding precision problem in tests, or if rpds should not be used in i686.
(full build.log at https://koji.fedoraproject.org/koji/taskinfo?taskID=105821360)
Thank you

orium commented

Hi. Thanks for reporting. The tests were written for 64 bit platforms and they fail on 32 bit computers. It's only the tests that have a problem, not the implementation itself. I'll publish a fix soon.

orium commented

Fix released in v1.0.1.

  • 🚀1