João Valverde wrote: > alex23 wrote: >> João Valverde <backu...@netcabo.pt> wrote: >> >>> Currently I don't have a strong need for this. >>> >> >> And clearly neither has anyone else, hence the absence from the >> stdlib. As others have pointed out, there are alternative approaches, >> and plenty of recipes on ActiveState, which seem to have scratched >> whatever itch there is for the data structure you're advocating. >> > > Propose such alternative then. There are none that offer the same > performance. At best they're workarounds. > > I don't care about recipes. That's called research. > > If people don't find it to be useful, that's fine. Surprising, but fine.
Python devs, based on my observation, tend to choose a data structure based on the interface and not its implementation. Binary Sorted Tree is an implementation, its interface can be a sorted dict (sorted key-value mapping) or a list (not the most natural interface for a tree). Basically, python already have all the common interfaces, i.e. list, set, and mapping/dict. Let's see it like this. In how many ways can a list be implemented? - Array (python's builtin list) - Linked List - Binary Tree - and the list goes on... All of them can expose their interface as list, but only array implementation is available as builtin. > And I don't have a need because I'm not using Python for my project. If > I wanted to I couldn't, without implementing myself or porting to Python > 3 a basic computer science data structure. > >> While Python's motto is "batteries included" I've always felt there >> was an implicit "but not the kitchen sink" following it. Just because >> something "could" be useful shouldn't be grounds for inclusion. That's >> what pypi & the recipes are for. Ideally, little should be created >> wholesale for the stdlib, what should be added are the existing 3rd >> party modules that have become so ubiquitous that their presence on >> any python platform is just expected. >> > > Agreed. -- http://mail.python.org/mailman/listinfo/python-list
