From 3377430769e4687f888c2e65005877b385359bb1 Mon Sep 17 00:00:00 2001
From: Brian Hicks <brian@brianthicks.com>
Date: Tue, 24 Aug 2021 17:15:44 -0500
Subject: [PATCH] calculate sizes in parallel

---
 BENCHMARKING.md | 22 ++++++++++++++++++++++
 src/main.rs     | 14 +++++++++-----
 2 files changed, 31 insertions(+), 5 deletions(-)

diff --git a/BENCHMARKING.md b/BENCHMARKING.md
index 033ead6..c29baa5 100644
--- a/BENCHMARKING.md
+++ b/BENCHMARKING.md
@@ -148,3 +148,25 @@ Summary
   './target/release/similar-sort benchmark < /usr/share/dict/words' ran
     1.74 ± 0.08 times faster than './result/bin/similar-sort define < /usr/share/dict/words'
 ```
+
+## Calculating sizes in parallel
+
+What if we calculated the size in parallel?
+Could we get it even faster?
+
+```
+hyperfine './result/bin/similar-sort define < /usr/share/dict/words' './target/release/similar-sort benchmark < /usr/share/dict/words'
+Benchmark #1: ./result/bin/similar-sort define < /usr/share/dict/words
+  Time (mean ± σ):     295.0 ms ±   5.6 ms    [User: 259.3 ms, System: 76.5 ms]
+  Range (min … max):   287.4 ms … 305.2 ms    10 runs
+
+Benchmark #2: ./target/release/similar-sort benchmark < /usr/share/dict/words
+  Time (mean ± σ):     153.5 ms ±   3.4 ms    [User: 143.0 ms, System: 11.0 ms]
+  Range (min … max):   147.5 ms … 163.0 ms    19 runs
+
+Summary
+  './target/release/similar-sort benchmark < /usr/share/dict/words' ran
+    1.92 ± 0.06 times faster than './result/bin/similar-sort define < /usr/share/dict/words'
+```
+
+Yep!
diff --git a/src/main.rs b/src/main.rs
index f57d537..6490164 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -25,17 +25,21 @@ fn try_main() -> Result<()> {
 
     let opts = Opts::from_args();
 
-    let mut lines: Vec<(usize, String)> = stdin()
+    let lines: Vec<String> = stdin()
         .lock()
         .lines()
-        .map(|line| line.map(|candidate| (levenshtein(&opts.target, &candidate), candidate)))
-        .collect::<io::Result<Vec<(usize, String)>>>()
+        .collect::<io::Result<Vec<String>>>()
         .context("could not read lines from stdin")?;
 
-    lines.par_sort_unstable_by_key(|x| x.0);
+    let mut distances: Vec<(usize, &String)> = lines
+        .iter()
+        .map(|candidate| (levenshtein(&opts.target, candidate), candidate))
+        .collect();
+
+    distances.par_sort_unstable_by_key(|x| x.0);
 
     let mut out = BufWriter::new(stdout());
-    for (_, candidate) in lines {
+    for (_, candidate) in distances {
         writeln!(out, "{}", candidate).context("could not write to stdout")?;
     }