Skip to content

Latest commit

 

History

History
34 lines (22 loc) · 1.09 KB

README.md

File metadata and controls

34 lines (22 loc) · 1.09 KB

RDF Dataset Canonicalization Benchmark

The benchmark system uses the same manifest setup as the test suite. This allows setting up both tests and benchmarks with one setup.

In the root dir run all test suite tests:

BENCHMARK=1 npm t

To run a particular dir or file:

BENCHMARK=1 TESTS=`pwd`/benchmark npm t
BENCHMARK=1 TESTS=`pwd`/benchmark/m2.jsonld npm t

The EARL env var can be used to output test results. The TEST_ENV env var is used to control data in the output. See the test script for details. Use the benchmark compare script in jsonld.js to compare different benchmark runs.

As an addition to the test suite, tests can be marked with boolean "skip" or "only" flags. The "only" support is a hack and requies the env var "ONLY" to be set:

ONLY=1 npm run benchmark
ONLY=1 TEST_DIR=./benchmark npm run benchmark

Tests are benchmarked with a matrix of {async/sync} {js/native} {x1/x10}.

To run large "block" benchmarks, run the builder script first:

node make-tests.js