There are several advantages to intrusive node-based data structures that I haven't seen stated in the article or the comments:
- the same object can move between containers with no allocation and no need for a dedicated complex API
- the same object can be part of multiple containers at once; particularly useful for intrusive binary trees, for indexing data with different criteria.
- the container can be fully polymorphic, no need for all elements to be the same dynamic type.
- no need for complex allocators, you can just store the objects as you see fit.
- the same object can move between containers with no allocation and no need for a dedicated complex API
- the same object can be part of multiple containers at once; particularly useful for intrusive binary trees, for indexing data with different criteria.
- the container can be fully polymorphic, no need for all elements to be the same dynamic type.
- no need for complex allocators, you can just store the objects as you see fit.