#! /bin/bash # # run-nrtest.sh - Runs numerical regression test # # Date Created: 10/16/2017 # # Author: Michael E. Tryby # US EPA - ORD/NRMRL # # Arguments: # 1 - relative path to location there test suite is staged # 2 - version/build identifier # run-nrtest() { return_value=0 test_suite_path=$2 benchmark_ver="220dev1" nrtest_execute_cmd="nrtest execute" test_app_path="apps/epanet-$3.json" tests="tests/examples tests/exeter tests/large tests/network_one tests/small tests/tanks tests/valves" test_output_path="benchmark/epanet-$3" nrtest_compare_cmd="nrtest compare" ref_output_path="benchmark/epanet-${benchmark_ver}" rtol_value=0.1 atol_value=0.0 # change current directory to test_suite cd ${test_suite_path} # clean test benchmark results rm -rf ${test_output_path} echo INFO: Creating test benchmark nrtest_command="${nrtest_execute_cmd} ${test_app_path} ${tests} -o ${test_output_path}" echo INFO: "$nrtest_command" return_value=$( $nrtest_command ) if [ $1 = 'true' ]; then echo echo INFO: Comparing test and ref benchmarks nrtest_command="${nrtest_compare_cmd} ${test_output_path} ${ref_output_path} --rtol ${rtol_value} --atol ${atol_value}" echo INFO: "$nrtest_command" return_value=$( $nrtest_command ) fi return $return_value } print_usage() { echo " " echo "run-nrtest.sh - generates artifacts for SUT and performes benchmark comparison " echo " " echo "options:" echo "-c, don't compare SUT and benchmark artifacts" echo "-t test_path relative path to location where test suite is staged" echo "-v version version/build identifier" } # Default option values compare='true' test_path='nrtestsuite' version='vXXX' while getopts ":ct:v:" flag; do case "${flag}" in c ) compare='false' ;; t ) test_path="${OPTARG}" ;; v ) version=${OPTARG} ;; \? ) print_usage exit 1 ;; esac done shift $(($OPTIND - 1)) # Invoke command run_command="run-nrtest ${compare} ${test_path} ${version}" echo INFO: "$run_command" $run_command