simon/jolt
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Persistent vector as a 32-way trie (#225)

Browse source 
The persistent vector was a flat Scheme vector with copy-on-write: every conj
copied the whole backing array, so building an n-element vector was O(n^2). On the
collections bench that's vec-sum building 7500 elements with 227MB of copies, 90%
of the bench's allocation.

Replace it with Clojure's PersistentVector — a 32-way trie plus a trailing tail
chunk. conj appends to the tail and, when it fills, path-copies it into the trie,
so conj is O(1) amortized and a linear build is O(n). nth/assoc/pop are
O(log32 n). make-pvec (build a trie from a flat vector) and pvec-v (materialize
back) stay as compatibility shims, so the ~14 callers that read the backing array
— all one-shot conversions in =, hash, seq, meta-copy, transients, the reader —
are untouched; only this file's internals change.

vec-sum 70ms/227MB -> 1ms/3MB; collections 10.4x -> 4.0x vs JVM, under the 5x
target. 5 new corpus rows plus boundary stress (level transitions at 32 and 1024,
pop-collapse, assoc at every index) cover the trie.

Co-authored-by: Yogthos <yogthos@gmail.com>
This commit is contained in:
Dmitri Sotnikov 2026-06-26 06:32:18 +00:00 • committed by GitHub
parent 93be40f3fe
commit 32ef74b9b0
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 151 additions and 30 deletions
Download patch file Download diff file Expand all files Collapse all files

5
test/chez/corpus.edn
View file

@ -3211,4 +3211,9 @@
{:suite "protocols / arity dispatch" :label "protocol method extended to a host type with an arg" :expected "\"S:hi:5\"" :actual "(do (defprotocol Q (mq [this a])) (extend-protocol Q java.lang.String (mq [this a] (str \"S:\" this \":\" a))) (mq \"hi\" 5))"}
{:suite "protocols / arity dispatch" :label "protocol method used as a value" :expected "[1 2]" :actual "(do (defprotocol Z (z [this])) (defrecord ZR [v] Z (z [_] v)) (mapv z [(->ZR 1) (->ZR 2)]))"}
{:suite "protocols / arity dispatch" :label "no-extra-arg method dispatches by type" :expected "[16 25]" :actual "(do (defprotocol Sh (ar [s])) (defrecord Sq [x] Sh (ar [_] (* x x))) (defrecord Sq2 [y] Sh (ar [_] (* y y))) [(ar (->Sq 4)) (ar (->Sq2 5))])"}
{:suite "vectors / large trie" :label "linear build equals range across paths" :expected "true" :actual "(= (vec (range 2000)) (reduce conj [] (range 2000)) (into [] (range 2000)))"}
{:suite "vectors / large trie" :label "nth across a multi-level vector" :expected "[0 1024 2000 2099]" :actual "(let [v (vec (range 2100))] [(nth v 0) (nth v 1024) (nth v 2000) (nth v 2099)])"}
{:suite "vectors / large trie" :label "assoc in a large vector" :expected "[1098 :x 1100]" :actual "(let [v (assoc (vec (range 1100)) 1099 :x)] [(nth v 1098) (nth v 1099) (count v)])"}
{:suite "vectors / large trie" :label "pop down across a level boundary" :expected "[1023 1022]" :actual "(let [p (pop (vec (range 1024)))] [(count p) (peek p)])"}
{:suite "vectors / large trie" :label "pop to empty then rebuild" :expected "[0 5]" :actual "(let [e (loop [v (vec (range 100))] (if (seq v) (recur (pop v)) v))] [(count e) (count (into e (range 5)))])"}
]