Returning "INCOMPLETE" instead a "ERROR"

[paolodragone]

Returning "INCOMPLETE" instead a "ERROR"

[paolodragone]

Hi @Dekker1, thank you for the tip, I'll make use of it!

[paolodragone]

Hi @ceandrade, thanks for reporting this. This happens because the error is raised at flattening time, so no error flag (=====ERROR=====) is printed on standard output (commonly printed at solving time), and so it's not caught during solution parsing. PyMzn looks for these errors before launching the full MiniZinc run by doing model checking and instance checking (i.e. running MiniZinc with --model-check-only and --instance-check-only). However, quite surprisingly, in your case MiniZinc does not fail during model nor instance checking, I'm not exactly sure why. I'll make a fix for this in PyMzn.

[ceandrade]

Yes, this is weird, I agree. BTW, my minizinc version is

$ minizinc --version
MiniZinc to FlatZinc converter, version 2.3.1, build 70205949
Copyright (C) 2014-2019 Monash University, NICTA, Data61

@paolodragone I will doubt check this and may open a ticket on minizinc repo.

Thanks for your fast reply!

[ceandrade]

FYI: MiniZinc/libminizinc#313

[paolodragone]

Thanks. Anyway, now it is solved in the master branch. I'll make a release soon.

[Dekker1]

@paolodragone The best way of ensuring the process finished successfully would be to check the exit code. When the exit code is non-zero then an error has occurred during execution.