Bug: Unable to re-run script after creating a class that inherits from base

[cslotboom]

When inheriting from base, I am unable to run a script twice without getting a value error my custom class. Looks like some persistent global registry object isn't getting properly refreshed on each run?

I'm creating the object below, and I get this error when running twice:

class PointGroup(Base, speckle_type="UserObject"):
    
    def __init__(self, points:list[Node], lines:list[Element2D] = None):
        self.points = points
        self.lines = lines

System Info

WIndows 11,
specklepy = 2.21

[linear]

CXPLA-130 Bug: Unable to re-run script after creating a class that inherits from base

[gjedlicska]

Hey @cslotboom, it would be nice to understand where you are running this script. The execution environment is probably trying to re-parse your code in a running environment.

I can only reproduce this in a plain python script by registering the same class in 2 different places:

from specklepy.objects.base import Base
from specklepy.objects.structural.geometry import Element2D, Node


class PointGroup(Base, speckle_type="UserObject"):
     
    def __init__(self, points:list[Node], lines:list[Element2D] = None):
        self.points = points
        self.lines = lines



class PointGroup(Base, speckle_type="UserObject"):
     
    def __init__(self, points:list[Node], lines:list[Element2D] = None):
        self.points = points
        self.lines = lines

produces the same error

Traceback (most recent call last):
  File "/home/gergojedlicska/Speckle/specklepy/example/register.py", line 13, in <module>
    class PointGroup(Base, speckle_type="UserObject"):
  File "/home/gergojedlicska/Speckle/specklepy/src/specklepy/objects/base.py", line 174, in __init_subclass__
    raise ValueError(
ValueError: The speckle_type: UserObject is already registered for type: PointGroup. Please choose a different type name.

so most probably this is what is happening in your scripting environment. Ideally this situation would be avoided, by only running the class definition once on startup, cause we have no extensive check to make sure the two type definitions match 1-1.

If you override the type definition with some differences, things can get weird unexpectedly