Testing & Miri

Run the full suite

cargo test-all

That’s an alias for cargo test --workspace --all-features — exercises every test in every crate under every optional feature (xsd, xslt, html, serde, tokio, network-resolver, full-encodings, c-abi) plus doctests, integration tests, and example builds.

Baseline: all green. A change that breaks the suite gets fixed in the same PR.

Miri

rustup toolchain install nightly
rustup +nightly component add miri
cargo +nightly miri setup
cargo +nightly miri test --workspace --exclude sup-xml-bench

Miri executes the suite one MIR instruction at a time, detecting every UB class — out-of-bounds reads/writes, use-after-free, invalid pointer arithmetic, aliasing violations, uninit reads. CI runs Miri on every PR.

sup-xml-bench is excluded because it links libxml2 via FFI, which Miri can’t interpret.

Fuzzing

cd crates/core/fuzz
cargo +nightly fuzz run parse_str -- -max_total_time=600

Corpora are in crates/core/fuzz/corpus/ — seeds drawn from W3C conformance, common feeds, and CVE reproducers from libxml2 history.

W3C conformance suite

The W3C suite is vendored at tests/w3c/ (the xmlconf-20130923 snapshot). The harness lives at crates/api/tests/w3c.rs and runs as a regular workspace test:

cargo test --release --test w3c -p sup-xml -- --nocapture

Expected output:

W3C XML Conformance: 2274 passed, 0 failed, 0 xfail (allow-listed),
                     21 skipped (of 2295 total)

The failing-IDs allowlist lives in the harness itself (KNOWN_FAILING_IDS, currently empty). A regression that drops the count below 2274 fails CI; a test that unexpectedly passes off the allowlist also fails, so the allowlist stays current.

Fast feedback during development

cargo check-all     # compiles everything, skips tests — fast
cargo test -p sup-xml-core   # just the core crate
cargo test --doc             # just doctests