Summary
Running cargo test --features zstd on a clean checkout of main produces two test failures:
failures:
algorithms::enhanced_ctw::tests::test_enhanced_ctw_basic
algorithms::multi_pass::tests::test_multi_pass_basic
test result: FAILED. 10 passed; 2 failed; 0 ignored
Both failures come from the same shape of assertion:
assert!(compressed.len() < test_data.len());
on inputs that are too small / too non-repetitive for the algorithm's framing overhead to pay for itself.
test_enhanced_ctw_basic
Input: b"Hello, world! This is a test of Enhanced CTW compression." (58 bytes). At this size the compressed stream header dominates. The roundtrip correctness assertion that precedes it passes; only the size assertion fails.
test_multi_pass_basic
Input: b"AAAA BBBB CCCC AAAA BBBB CCCC".repeat(10) (290 bytes). MultiPassCompressor::apply_pattern_replacement compares data[pos..pos+4] with data[pos+4..pos+8] (immediately-adjacent 4-byte blocks), which this input never contains. With best_ratio initialized to 1.0, the first pass shows no improvement and the method returns the original bytes unchanged, so compressed.len() == test_data.len() and the assertion fails. There is also no roundtrip assertion at all in this test.
Proposed fix
Small, scoped test-only change:
test_enhanced_ctw_basic: use a larger, more compressible corpus (e.g. repeat the sentence enough times that compression is plausible). Keep the existing roundtrip assertion.test_multi_pass_basic: switch to input with immediately-repeated 4-byte blocks (e.g. b"AAAAAAAABBBBBBBBCCCCCCCCDDDDDDDD".repeat(N)) so apply_pattern_replacement can actually compress, and drop the assertion or guard it behind a roundtrip once decompress is implemented. For now, assert a ratio on the pass that produces a real output.
Happy to send a PR.
Summary
Running
cargo test --features zstdon a clean checkout ofmainproduces two test failures:Both failures come from the same shape of assertion:
on inputs that are too small / too non-repetitive for the algorithm's framing overhead to pay for itself.
test_enhanced_ctw_basicInput:
b"Hello, world! This is a test of Enhanced CTW compression."(58 bytes). At this size the compressed stream header dominates. The roundtrip correctness assertion that precedes it passes; only the size assertion fails.test_multi_pass_basicInput:
b"AAAA BBBB CCCC AAAA BBBB CCCC".repeat(10)(290 bytes).MultiPassCompressor::apply_pattern_replacementcomparesdata[pos..pos+4]withdata[pos+4..pos+8](immediately-adjacent 4-byte blocks), which this input never contains. Withbest_ratioinitialized to1.0, the first pass shows no improvement and the method returns the original bytes unchanged, socompressed.len() == test_data.len()and the assertion fails. There is also no roundtrip assertion at all in this test.Proposed fix
Small, scoped test-only change:
test_enhanced_ctw_basic: use a larger, more compressible corpus (e.g. repeat the sentence enough times that compression is plausible). Keep the existing roundtrip assertion.test_multi_pass_basic: switch to input with immediately-repeated 4-byte blocks (e.g.b"AAAAAAAABBBBBBBBCCCCCCCCDDDDDDDD".repeat(N)) soapply_pattern_replacementcan actually compress, and drop the assertion or guard it behind a roundtrip oncedecompressis implemented. For now, assert a ratio on the pass that produces a real output.Happy to send a PR.