Data representation in C : Linked structs vs Big structs

Question

While looking at someone's code i was struck by a way of representing data which, in a C context, seemed strange to me :

This person needed to represent different kind of geometries for a raytracer, each geometries while being different had similarities in their structures. So this person decided to represent the geometries through more or less abstract structs linked together through pointers.

s_entities -> s_scene_object -> s_sphere
s_entities -> s_scene_object -> s_plane
s_entities -> s_pointlight

i was wondering why one would decompose its data in C through abstraction layers instead of having an all-encompassing struct for each kind of data.

One explanation would be to reduce overhead if you have functions that operate on a certain layer of abstration of your data. right ?

it seems to me like remnants of OOP habits.

Answer 1

You say "remnants of OOP habits" as if that's something that's obviously not OK in C. I'd say you're wrong, there are rather large collections of very heavily object-oriented code in C. It works.

Object-orientedness is a way to arrange and model things, it's not tied to programming languages. C is low-level enough that you can do all sorts of programming paradigms in it; you'll often have to implement much of the paradigm yourself since there is little support from the language, but OO is pretty straight-forward.

It's not very clear from your example, but a more typical way to model things like that in C is by embeddeding the more basic data structure (superclass) inside the more specialized, utilizing the fact that the first member in a C struct is at offset 0 from the struct itself:

typedef struct {
  float x, y, z;
} Shape;

typedef struct {
  Shape shape;
  float radius;
} Sphere;