Expansion of structure of blender data base

Created by: Durman

Problem statement

As far as I know there are two ways to adding custom values to Blender objects. First is similar to working with Python dictionary. Second is creating attributes of object by assign to them a Blender property.

D.objects['Cube']['my_var'] = 1

bpy.types.Object.my_var = bpy.props.IntigerProperty()
D.objects['Cube'].my_var = 1

Viewer BMesh looks like using first way. While I worked with material node I found second way more useful because it allows to keep link to other objects like ths:

bpy.types.Object.my_mat = bpy.props.PointerProperty(name='my_mat', type=bpy.types.Material)
mat = D.materials['Material']
D.objects['Cube'].my_mat = mat

So I created quite complicated data structure and according this I have two questions:

Data stored in Material internal type of data of Blender:
Material - (type=ID)
  └ sv_props - (type=Group)
    ├ sv_created - (type=Bool)
    └ children_branch - (type=Collection)
      ├ nodes_data - (type=Collection)
      │ └ required_number - (type=Int)
      └ children - (type=Collection)
        └ mat - (type=ID)

• Why Sverchok does not use second way more wide. Is it because this way have less stability or there was no use in using such way before?
• And If I'm going to develop this approach further where should I store the code of such data structure?