From 5aef0c488fb93c3a5384ed17129544ce998e7c8a Mon Sep 17 00:00:00 2001
From: Simon Ochsenreither <simon@ochsenreither.de>
Date: Tue, 8 Aug 2023 01:57:38 +0200
Subject: [PATCH] stdlib: better `Hash` for `Float64`/`Float32`

fixes #90
---
 dora/stdlib/primitives.dora             | 12 ++++++++++--
 tests/float/float32-hash.dora           | 12 ++++++++++++
 tests/float/float64-hash.dora           | 12 ++++++++++++
 tests/num-hash.dora                     |  3 ---
 tests/stdlib/hashmap-contains-zero.dora |  8 ++++----
 tests/stdlib/hashset-contains-zero.dora |  4 ++--
 6 files changed, 40 insertions(+), 11 deletions(-)
 create mode 100644 tests/float/float32-hash.dora
 create mode 100644 tests/float/float64-hash.dora

diff --git a/dora/stdlib/primitives.dora b/dora/stdlib/primitives.dora
index 5cfb600d..07ffb822 100644
--- a/dora/stdlib/primitives.dora
+++ b/dora/stdlib/primitives.dora
@@ -350,7 +350,10 @@ impl Float32 {
   @pub @internal fun isNan(): Bool;
   @pub @internal fun sqrt(): Float32;
 
-  @pub fun hash(): Int32 = self.asInt32();
+  @pub fun hash(): Int32 =
+    // discard the sign, such that 0.0 and -0.0 hash the same, at the cost of all positive numbers
+    // and their negative counterparts hashing the same, but keeping the computation branch-free.
+    self.asInt32() & 0x7FFFFFFFi32;
 
   // should be lets, not funs
   @pub @static fun bits(): Int32 = 32i32;
@@ -405,7 +408,12 @@ impl Float64 {
   @pub @internal fun isNan(): Bool;
   @pub @internal fun sqrt(): Float64;
 
-  @pub fun hash(): Int32 = self.asInt64().toInt32();
+  @pub fun hash(): Int32 {
+    // discard the sign, such that 0.0 and -0.0 hash the same, at the cost of all positive numbers
+    // and their negative counterparts hashing the same, but keeping the computation branch-free.
+    let hash = self.asInt64() & 0x7FFFFFFFFFFFFFFF;
+    (hash ^ hash.shiftRight(31i32)).toInt32()
+  }
 
   // should be lets, not funs
   @pub @static fun bits(): Int32 = 64i32;
diff --git a/tests/float/float32-hash.dora b/tests/float/float32-hash.dora
new file mode 100644
index 00000000..e9d8a29a
--- /dev/null
+++ b/tests/float/float32-hash.dora
@@ -0,0 +1,12 @@
+fun main(): Unit {
+  assert(  0.0f32 .hash() == 0i32);
+  assert((-0.0f32).hash() == 0i32);
+
+  assert(  1.0f32 .hash() == 1065353216i32);
+  assert((-1.0f32).hash() == 1065353216i32);
+
+  assert(Float32::infinityPositive().hash() == 2139095040i32);
+  assert(Float32::infinityNegative().hash() == 2139095040i32);
+
+  assert(Float32::notANumber().hash() == 2143289344i32);
+}
diff --git a/tests/float/float64-hash.dora b/tests/float/float64-hash.dora
new file mode 100644
index 00000000..3ecdd9cb
--- /dev/null
+++ b/tests/float/float64-hash.dora
@@ -0,0 +1,12 @@
+fun main(): Unit {
+  assert(  0.0 .hash() == 0i32);
+  assert((-0.0).hash() == 0i32);
+
+  assert(  1.0 .hash() == 2145386496i32);
+  assert((-1.0).hash() == 2145386496i32);
+
+  assert(Float64::infinityPositive().hash() == -2097152i32);
+  assert(Float64::infinityNegative().hash() == -2097152i32);
+
+  assert(Float64::notANumber().hash() == -1048576i32);
+}
diff --git a/tests/num-hash.dora b/tests/num-hash.dora
index dfe009af..b332e7c6 100644
--- a/tests/num-hash.dora
+++ b/tests/num-hash.dora
@@ -4,7 +4,4 @@ fun main(): Unit {
   assert(1.toUInt8().hash() == 1i32);
   assert(1.hash() == 1i32);
   assert(1i64.hash() == 1i32);
-  assert(1.0f32.hash() == 1065353216i32);
-  // double needs a better hash implementation
-  assert(1.0.hash() == 0i32);
 }
diff --git a/tests/stdlib/hashmap-contains-zero.dora b/tests/stdlib/hashmap-contains-zero.dora
index b7ed8724..eb648bbc 100644
--- a/tests/stdlib/hashmap-contains-zero.dora
+++ b/tests/stdlib/hashmap-contains-zero.dora
@@ -9,8 +9,8 @@ fun float64(): Unit {
   assert(map.size() == 1);
   assert(map.contains(0.0));
   assert(map.get(0.0).getOrPanic() == "a");
-  assert(map.contains(-0.0)); // FIXME: this only works by coincidence, because the hash function sucks
-  assert(map.get(-0.0).getOrPanic() == "a"); // FIXME: this only works by coincidence, because the hash function sucks
+  assert(map.contains(-0.0));
+  assert(map.get(-0.0).getOrPanic() == "a");
 }
 
 fun float32(): Unit {
@@ -19,6 +19,6 @@ fun float32(): Unit {
   assert(map.size() == 1);
   assert(map.contains(0.0f32));
   assert(map.get(0.0f32).getOrPanic() == "a");
-  assert(map.contains(-0.0f32).not()); // FIXME: should be true instead
-  // assert(map.get(-0.0f32).getOrPanic() == "a"); // FIXME: should be true instead
+  assert(map.contains(-0.0f32));
+  assert(map.get(-0.0f32).getOrPanic() == "a");
 }
diff --git a/tests/stdlib/hashset-contains-zero.dora b/tests/stdlib/hashset-contains-zero.dora
index 581ef099..09cf48de 100644
--- a/tests/stdlib/hashset-contains-zero.dora
+++ b/tests/stdlib/hashset-contains-zero.dora
@@ -8,7 +8,7 @@ fun float64(): Unit {
 
   assert(set.size() == 1);
   assert(set.contains(0.0));
-  assert(set.contains(-0.0)); // FIXME: this only works by coincidence, because the hash function sucks
+  assert(set.contains(-0.0));
 }
 
 fun float32(): Unit {
@@ -16,5 +16,5 @@ fun float32(): Unit {
 
   assert(set.size() == 1);
   assert(set.contains(0.0f32));
-  assert(set.contains(-0.0f32).not()); // FIXME: should be true instead
+  assert(set.contains(-0.0f32));
 }