pyiron_base
pyiron_base copied to clipboard
Default job name of restarted simulations
I'm currently working on Damask where restarting is part of job routines. In the current setting, the default job name is job_restart
if the job name of the old job is job
. This causes a bit of problems when the job has to be let's say run 10 times, which will be something like job_restart_restart_restart_..._restart
. I'd like to change it but I also wanted to hear your opinions
- Possibility 0: Leave it like this
- Possibility 1: Replace it by
rerun_{index}
(orrestart
, butrerun
is shorter)- This makes it difficult to bundle all the simulations, because the first run won't contain
rerun
, so if we use a for loop it's gonna be likeall_jobs = [pr.load("job")] + [pr.load(f"job_rerun_{counter}") for counter in range(5)]
(which is still WAY easier than the current setting). - We could potentially check if
job_restart
exists and make a warning for the backward compatibility
- This makes it difficult to bundle all the simulations, because the first run won't contain
- Possibility 2: Create a new entry in the job table for the counter
- It's probably a nightmare for the backward compatibility
- We wouldn't have a problem with user accidentally using
rerun
orrestart
in the job name (because in the option above we would have to parsererun
orrestart
)