Fractal Tree to create fast, snapshottable, massively scalable databases

github.com

21 points·jgrodziski··5 comments

Previous discussion: https://news.ycombinator.com/item?id=12229906

I don't understand the difference between this data structure [0] and buffered B-trees, as described in "Lower Bounds for External Memory Dictionaries", by Gerth Stølting Brodal and Rolf Fagerberg [1], or in "On External Memory Graph Traversal" by Adam L. Buchsbaum, Michael Goldwasser, Suresh Venkatasubramanian, and Jeffery R. Westbrook [2]. Granted, this adds in path-copying, but I'm not sure what new innovation this has that merits the description "Hitchhiker trees are a newly invented (by @dgrnbrg) datastructure".

0: https://github.com/datacrypt-project/hitchhiker-tree/blob/ma...

1: http://www.cs.au.dk/~gerth/papers/alcomft-tr-03-75.pdf

2: http://euler.slu.edu/~goldwasser/publications/SODA2000.pdf

Maybe add 'in clojure' to the title?

I am always surprised how people care about the language of implementation so much. The data structure is interesting by itself. Why should we care what this is built with?

Because that may mean that it is not usable for you. Not everybody is interested in the theory of things, and once you implement you tie things down to a specific ecosystem.

Adding 'in clojure' does not detract from the info already in the title but makes it more informative. This particular implementation is also quite dependent (as far as I can see) on bits and pieces that clojure supplies natively but that may be hard to implement in other languages.

Reply on news.ycombinator.com