Relevance of global import in submodules

[esnguyenvan]

For example in the component compute VT :

import openmdao.api as om

import fastoad.api as oad
from fastoad.models.geometry.geom_components.vt.components import ComputeVTChords
from fastoad.models.geometry.geom_components.vt.components import ComputeVTClalpha
from fastoad.models.geometry.geom_components.vt.components import ComputeVTDistance
from fastoad.models.geometry.geom_components.vt.components import ComputeVTMAC
from fastoad.models.geometry.geom_components.vt.components import ComputeVTSweep
from fastoad.module_management.service_registry import RegisterSubmodel
from ...constants import SERVICE_FUSELAGE_CNBETA, SERVICE_VERTICAL_TAIL_GEOMETRY

What are the reasons for using global imports while the lower modules ComputeVTChords, etc... are readily available through a relative import?

I imagine a case where someone would like to register a new VT module. Personally, I would start by copying the whole 'vt' folder and register it as a new submodule. I would then modify the lower modules ComputeVTChords, etc... and finally I will have to change from global imports to relative imports for the changes to actually take place. This last step may a be source of confusion if ones expects from a custom module to strictly use the submodules it contains.

[christophe-david]

Agreed. Relative imports are the right thing to do here.

Same applies for wing and horizontal tail.

[esnguyenvan]

I can do the modifications at the same time than takeoff enhancements, which I'm currently working on

[christophe-david]

Actually, I realize that incoming changes in code organisation should make this change unavoidable anyway, so don't bother too much with that.

[christophe-david]

Models have been moved to FAST-OAD_CS25 repo, and imports have been modified BTW.