зеркало из https://github.com/microsoft/QBASHER.git
d71c390e4e | ||
---|---|---|
.. | ||
Fuzz_7_100000_7200_0.q | ||
README.txt | ||
emulated_log.q | ||
emulated_log_1k.q | ||
emulated_log_10k.q | ||
emulated_log_100k.q | ||
emulated_log_100k_autopartials.q | ||
emulated_log_four_full_words_10k.q | ||
emulated_log_four_words_with_operators.q |
README.txt
Notes on the Use and Origin of the query sets in this directory. ---------------------------------------------------------------- 12 Dec 2017 Use: The query sets in this directory are relied upon by the build-test scripts in ../scripts. By running large query sets with different QBASHER options we are able to achieve reasonable confidence that changes to the code base have not seriously slowed down latency and throughput, and have not introduced segfaults or memory leaks. Fuzz Origin: The Fuzz query set was generated using the generate_fuzz_queries program in ../src/generate_fuzz_queries. It consists of randomly generated ASCII bytes, including control characters and punctuation. The aim is to reduce the chance that QBASHER will crash when highly garbled input is received. Emulated Log Origin: The emulated_log.q file was emulated using the queryLogEmulator program distributed as part of the SynthaCorpus project using a de-identified real query log. The -obfuscate option was used, and the target corpus was ../test_data/wikipedia_titles. The other query sets were derived from emulated_log.q using ../scripts/make_derivative_query_logs.pl. Note that ../scripts/qbash_classifier_timing_check.pl concatenates 100 copies of emulated_log.q in order to create a large batch.