It's just a toy pet-project written for fun, with no commitment to improve or mantain it at all. The code is poorly tested, contains no important optimizations (e.g. shortcut to tail node, transients) and probably isn't production-ready.
The absence of an efficient Clojure-like purelly functional vector in the Guile Scheme standard library and in the Guile ecosystem at all1 has always been a pain point for me. However, the idea of a bit-partitioned vector trie is very beautiful and relatively simple. So once I decided to implement it myself.
See the manual and the documentation strings in pvector.scm.
Reads from a random index, average time in seconds (lesser is better), with various element count. Performance of persistent vector, scheme arrays (SRFI-43), vlists, vhashes, linked lists. Logarithmic scale on both axis. Vlist is slightly better than pvector, but it doesn't have a setter for the random index. Vector is imperative.
Reads from a random index, average time in seconds (lesser is better), with various element count. Performance of persistent vector, scheme arrays (SRFI-43), vlists, vhashes (linked lists are unacceptably slow with element count involved). Logarithmic scale on data size axis. Again, vlists can't update a random element in the middle of the data structure.
Vhash is better than pvector, but see the data for random reads.
- Persistent vector implementation in Cloure by Rich Hickey: https://github.com/clojure/clojure/blob/master/test/clojure/test_clojure/vectors.clj
- Persistent vector implementation in Racket by Alexis King: https://github.com/lexi-lambda/racket-pvector/tree/master
- Blog post series «Understanding Clojure's Persistent Vectors» by Jean Niklas L'Orange:
- https://hypirion.com/musings/understanding-persistent-vector-pt-1
- https://hypirion.com/musings/understanding-persistent-vector-pt-2
- https://hypirion.com/musings/understanding-persistent-vector-pt-3
- https://hypirion.com/musings/understanding-clojure-transients
- https://hypirion.com/musings/persistent-vector-performance
- https://hypirion.com/musings/persistent-vector-performance-summarised
- A talk «Postmodern immutable data structures» by Juan Pedro Bolivar Puente at CppCon 2017: https://www.youtube.com/watch?v=sPhpelUfu8Q
AGPL v3
Footnotes
-
Of course, Lokke vector from the Lokke project and fectors should be mentioned; but I don't understand how to use Lokke vector outside of Lokke environment (and even is it possible at all or no), and the latter project looks abandoned more than decade ago. ↩