By chance I discovered that a completely blank input file can be run without any visible warning/crash.
To test it, just create a empty *.inp file and run it. The code will write (in a loop) the following message in the *.out “Unexpected end of input: START assumed”.
Indeed I discovered this issue since my quota was completely full after a few hours.
I would suggest that the code stops running when the card START is not find, instead of assuming it and tentatively run the input. It is a very unlikely situation that one run a blank input file (it happened to me because the *.inp was manipulated by a script, which accidentally created it), but in principle can happen.
Thanks for the report! Will take a look.
PS: the logics will be slightly different, since it is still legitimate to have an input file with just the STOP card (nice debugging trick to make sure that the code at least starts).
Dear @cesc ,
Thanks for the reply. True about the use of the card STOP for debugging -> in this case, shouldn’t the code look for STOP if START is not present in the .inp? I.e. without both cards you cannot either run/debug the simulation.
Thanks for having a look and for implementing the best logic to solve the issue.
It does. With “at least starts” only preliminary aspects were implied: ensure the binary was properly linked and executes, check paths, make sure random seed is picked up, etc.
Thanks once again for the report!