Closed jlegrand62 closed 3 years ago
Suggestion for improve: ensure that 'check_pipe.sh' store and keep the results of a scan for the two pipes (ml and geom) instead of overwriting.
Could you please explain a little bit more about the goal of this task @jlegrand62 and @fabfabBesnard ?
How about fixing the Clean
command since it seems to be the root problem?
The starting point was indeed to overcome the troubles with the Clean task. However it escalated quickly to adding more options to also use the files check*_pipe.py for local testing purposes rather than only for CI test operations...
So now you can quickly test changes locally by using check*_pipe.py --tmp
.
I admit that it may look redundant with the previous work we did with romi_run_task.py
.
But notes that there are other options that also helps to test a specific task (other than AnglesAndInternodes
) and dataset (other than the real or virtual plant).
Also, since we might evolve to a new data model, the Clean task will become obsolete... So I preferred not to waste time on it. However if we want to keep the original romidb implementation 'alive', we will have to correct the Clean task!
Suggestion for improve: ensure that 'check_pipe.sh' store and keep the results of a scan for the two pipes (ml and geom) instead of overwriting.
If you use the --tmp
option, it will create a dataset name with a $date_$time_$dataset
format like:
2021-01-27_11-48-15_GEOM_real_plant
2021-01-27_11-50-48_ML_real_plant
Close #150
Notes that as of now, the pipeline is not tested on the virtual plant anymore (anyway it fails according to @naittaleb).