From 786816c55fedcb4f40e4556c02461b5c99081950 Mon Sep 17 00:00:00 2001
From: Kenny Daniel <platypii@gmail.com>
Date: Sat, 19 Apr 2025 01:15:13 -0700
Subject: [PATCH] Set dictionary threshold to 2

---
 src/column.js | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/src/column.js b/src/column.js
index 7320c3e..3197448 100644
--- a/src/column.js
+++ b/src/column.js
@@ -25,7 +25,7 @@ export function writeColumn(writer, schemaPath, values, compressed, stats) {
   const statistics = stats ? getStatistics(values) : undefined
 
   // dictionary encoding
-  let dictionary_page_offset = undefined
+  let dictionary_page_offset
   let data_page_offset = BigInt(writer.offset)
   /** @type {DecodedArray | undefined} */
   const dictionary = useDictionary(values, type)
@@ -81,11 +81,9 @@ function useDictionary(values, type) {
   const unique = new Set(values)
   unique.delete(undefined)
   unique.delete(null)
-  if (values.length > 10 && values.length / unique.size > 0.1) {
-    if (unique.size < values.length) {
-      // TODO: sort by frequency
-      return Array.from(unique)
-    }
+  if (values.length / unique.size > 2) {
+    // TODO: sort by frequency
+    return Array.from(unique)
   }
 }