Trees

[Diez B. Roggisch]

> Writing a *simple* node class is easy, but a full-featured one that
> supports things like comparison and easy iteration is a bit more work. So
> various people write partial implementations with only the features they
> need, and they all end up being incompatible. So beyond being able to use
> it, the other thing is that having an implementation in the standard
> library acts as a de facto interface standard, making it easier to write
> common code to operate on tree structures.

Hm. I disagree on that. Some things make only sense for binary trees, others
need explicit uplinking to the parent node and so on. Some want accessor
methods, others have certain operator overloadings used for a specific
functionality. So  in the end, it either becomes a monstrosity or is again
not usable. But that's only my HO.

> All it would take to make it happen is a PEP, an implementation and a
> champion with some persuasive ability :)

Go wild :)
-- 
Regards,

Diez B. Roggisch