bench-shex

bench-shex

Covers the ShEx pipeline. Cases are vendored under bench-shex/corpus/{small,large}.zip, each containing a manifest.toml and its case files. On first use the zips are extracted into target/bench-shex-corpus/{small,large}/ (respecting CARGO_TARGET_TMPDIR / CARGO_TARGET_DIR).

• small: minimal cases from shexTest (single-shape schemas, tiny RDF). Times in the µs–ms range.
• large: real-world schemas (FHIR R5). Times in the ms–s range.

Each Criterion benchmark splits into two groups (<stage>_small, <stage>_large) so plots in each group share a comparable scale.

Stages

Bench	What it measures
parse	ShExC text to AST
compile	AST to IR
validator_init	IR to Validator (negation-cycle check)
validate	Run validation
end_to_end	Full rudof_lib flow: load → validate → serialize
validate_precompiled	Same case, two variants: from_source (ShExC) vs from_precompiled (.rsir cache)

Running

All benches:

cargo bench -p bench-shex

A specific stage:

cargo bench -p bench-shex --bench validate

A subset by group or case:

cargo bench -p bench-shex --bench compile -- compile_small
cargo bench -p bench-shex --bench validate -- 'large/fhir-r5'

HTML reports are written to target/criterion/; open target/criterion/report/index.html for the ov