this repo has no description
1# MiniZinc benchmarking automation
2
3Allows checking of MiniZinc solutions by configurable checker profiles. The
4solutions can be input from previous logs or produced by a chosen solver
5profile, for 1 or several instances, with result comparison and some
6ranking. New solutions' summary logs, detailed stdout/err outputs, and
7statistics are saved in subfolder mzn-test/LOGS, /OUTPUTS, and /STATS, resp.
8
9## USAGE EXAMPLES
10
11(1) "mzn-test.py model.mzn data.dzn [--checkDZN stdout.txt
12[--checkStderr stderr.txt]] [--chkPrf MINIZINC-CHK --chkPrf FZN-GECODE-CHK]
13[--tCheck 15] [--addSolverOption "--fzn-flags '-D fPureCircuit=true'"]" :::
14check the instance's solutions, optionally reading them from a DZN-formatted
15file (otherwise solving first), optionally overriding default checker list
16etc.
17
18(2) "mzn-test.py --slvPrf MZN-CPLEX -t 300 -l instList1.txt -l
19instList2.txt --name CPLEXTest_003 --result newLog00.json prevLog1.json
20prevLog2.json --failed failLog.json" ::: solve instances using the specified
21solver profile and wall time limit 300 seconds. The instances are taken from
22the list files. The test is aliased CPLEXTest_003. Results are saved to
23newLog00.json and compared/ranked to those in prevLog's. (Probably) incorrect
24solutions are saved to failLog.json.
25
26(3) "mzn-test.py [-l instList1.txt] -c
27prevLog1.json -c prevLog2.json [--runAndCmp]" ::: compare existing logs,
28optionally limited to the given instances, optionally running new tests. USE
29SINGLE QUOTES ONLY INSIDE ARGUMENTS PASSED TO THE BACKENDS when running
30backends through shell.
31