I would start with a tree primitive. That’s difficult enough, with variations on branching factor, whether the tree is self-balancing, whether it is immutable, etc)
I guess/expect that will have to expose the implementation detail of the branching factor (yes, a binary tree is a special-case of a https://en.wikipedia.org/wiki/B-tree for B = 2, but algorithms tend to be different)
On top of that, build a library of tree traversal routines (breadth first, depth first with a few variants), allowing the user to specify code that decides, on each node visited, whether to navigate further down and whether to stop the traversal.
Now, whether the language proper needs a generic way to specify tree traversals? I doubt it, and doubt even more that it is possible too that in a way so that a) most graph traversal algorithms can use the primitive and b) the result is easier to use/looks significantly nicer than just calling those routines.
I guess/expect that will have to expose the implementation detail of the branching factor (yes, a binary tree is a special-case of a https://en.wikipedia.org/wiki/B-tree for B = 2, but algorithms tend to be different)
On top of that, build a library of tree traversal routines (breadth first, depth first with a few variants), allowing the user to specify code that decides, on each node visited, whether to navigate further down and whether to stop the traversal.
Now, whether the language proper needs a generic way to specify tree traversals? I doubt it, and doubt even more that it is possible too that in a way so that a) most graph traversal algorithms can use the primitive and b) the result is easier to use/looks significantly nicer than just calling those routines.