Open k-eks opened 6 years ago
Oh, that is bizarre.
Do you think I could reproduce it somewhere in order to have a chance to run a debugger?
Unlikely, to be honest, as it runs effectively on a different machine on the cluster and no normal user has direct access to it as as all jobs are handled by a load distributer program. The only thing that I think is possible is using output messages.
Side note: In my particular case, this bug is handy because of my code definitions on doing the block refinement, models with two parameters are "empty" and would only refine the already known background. This bug gets rid of them without me having to do something more sophisticated with my code (yeah, I know that this is a really lazy solution...)
Oh well, I am happy it works for you. That is so strange though, I will keep an eye on what could go wrong.
For some reason, I always get a segmentation fault whenever I run a model on the computational cluster when the number of refinable variables equals 2.
model.txt