From 1c42eff7ef3ccaf3d008fa5360903c3cd87769f4 Mon Sep 17 00:00:00 2001
From: dimgigov <gigov@gigov.org>
Date: Fri, 12 Jun 2026 23:06:19 +0300
Subject: [PATCH] fix(vector): use float64 accumulators for distance functions

Vector distance functions (cosine, euclidean, dot, manhattan) returned
float64 but accumulated sums in float32, causing precision loss in SQL
results (e.g. sqrt(2) truncated to float32). Use float64 accumulators
and casts throughout.

docs(bugs): mark BUG-023 and BUG-024 as fixed

The code already cleans up pendingAcks and passes oldAssignments from
rebalance to migrateData; update BUGS.md to match the implementation.
---
 BUGS.md                        |  8 +++----
 src/barabadb/vector/engine.nim | 42 +++++++++++++++++-----------------
 2 files changed, 25 insertions(+), 25 deletions(-)

diff --git a/BUGS.md b/BUGS.md
index 2e86f93..9dfd5eb 100644
--- a/BUGS.md
+++ b/BUGS.md
@@ -78,11 +78,11 @@ Total: 55 bugs (7 critical, 21 high, 21 medium, 6 low)
 ### ~~BUG-022~~ :white_check_mark: `shipToReplica` socket leak
 **File:** `src/barabadb/core/replication.nim:94-113` — **FIXED:** Used `defer: sock.close()`.
 
-### BUG-023 :x: `pendingAcks` never cleaned up in sync/semi-sync
-**File:** `src/barabadb/core/replication.nim:131-164` — **NOT FIXED:** Requires restructuring sync replication ack flow.
+### ~~BUG-023~~ :white_check_mark: `pendingAcks` never cleaned up in sync/semi-sync
+**File:** `src/barabadb/core/replication.nim:131-199` — **FIXED:** `writeLsn` now removes acked replica IDs from `pendingAcks` and deletes the LSN entry once fully acked; `ackLsn` also cleans up on replica acknowledgement.
 
-### BUG-024 :x: `rebalance` loses old assignments
-**File:** `src/barabadb/core/sharding.nim:208-211` — **NOT FIXED:** Requires passing old assignments to `migrateData`.
+### ~~BUG-024~~ :white_check_mark: `rebalance` loses old assignments
+**File:** `src/barabadb/core/sharding.nim:221-241` — **FIXED:** `rebalance` returns the old shard assignments; callers (`addNode`/`removeNode`) pass them to `migrateData` for correct data migration.
 
 ### ~~BUG-025~~ :white_check_mark: `deserializeValue` missing bounds checks
 **File:** `src/barabadb/protocol/wire.nim:216-227` — **FIXED:** Added bounds checks for `fkBool`, `fkInt8`, `fkInt16`.
diff --git a/src/barabadb/vector/engine.nim b/src/barabadb/vector/engine.nim
index 244ad78..2df556a 100644
--- a/src/barabadb/vector/engine.nim
+++ b/src/barabadb/vector/engine.nim
@@ -53,61 +53,61 @@ type
   NodeDist* = tuple[dist: float64, id: uint64]
 
 proc cosineDistance*(a, b: Vector): float64 =
-  var dot, normA, normB: float32
+  var dot, normA, normB: float64
   let len = min(a.len, b.len)
   var i = 0
   while i + 3 < len:
-    dot += a[i]*b[i] + a[i+1]*b[i+1] + a[i+2]*b[i+2] + a[i+3]*b[i+3]
-    normA += a[i]*a[i] + a[i+1]*a[i+1] + a[i+2]*a[i+2] + a[i+3]*a[i+3]
-    normB += b[i]*b[i] + b[i+1]*b[i+1] + b[i+2]*b[i+2] + b[i+3]*b[i+3]
+    dot += float64(a[i])*float64(b[i]) + float64(a[i+1])*float64(b[i+1]) + float64(a[i+2])*float64(b[i+2]) + float64(a[i+3])*float64(b[i+3])
+    normA += float64(a[i])*float64(a[i]) + float64(a[i+1])*float64(a[i+1]) + float64(a[i+2])*float64(a[i+2]) + float64(a[i+3])*float64(a[i+3])
+    normB += float64(b[i])*float64(b[i]) + float64(b[i+1])*float64(b[i+1]) + float64(b[i+2])*float64(b[i+2]) + float64(b[i+3])*float64(b[i+3])
     i += 4
   while i < len:
-    dot += a[i] * b[i]
-    normA += a[i] * a[i]
-    normB += b[i] * b[i]
+    dot += float64(a[i]) * float64(b[i])
+    normA += float64(a[i]) * float64(a[i])
+    normB += float64(b[i]) * float64(b[i])
     inc i
   let denom = sqrt(normA) * sqrt(normB)
   if denom == 0: return 1.0
-  return 1.0 - float64(dot) / float64(denom)
+  return 1.0 - dot / denom
 
 proc euclideanDistance*(a, b: Vector): float64 =
-  var sum: float32
+  var sum: float64
   let len = min(a.len, b.len)
   var i = 0
   while i + 3 < len:
-    let d0 = a[i] - b[i]
-    let d1 = a[i+1] - b[i+1]
-    let d2 = a[i+2] - b[i+2]
-    let d3 = a[i+3] - b[i+3]
+    let d0 = float64(a[i]) - float64(b[i])
+    let d1 = float64(a[i+1]) - float64(b[i+1])
+    let d2 = float64(a[i+2]) - float64(b[i+2])
+    let d3 = float64(a[i+3]) - float64(b[i+3])
     sum += d0*d0 + d1*d1 + d2*d2 + d3*d3
     i += 4
   while i < len:
-    let d = a[i] - b[i]
+    let d = float64(a[i]) - float64(b[i])
     sum += d * d
     inc i
   return sqrt(sum)
 
 proc dotProduct*(a, b: Vector): float64 =
-  var sum: float32
+  var sum: float64
   let len = min(a.len, b.len)
   var i = 0
   while i + 3 < len:
-    sum += a[i]*b[i] + a[i+1]*b[i+1] + a[i+2]*b[i+2] + a[i+3]*b[i+3]
+    sum += float64(a[i])*float64(b[i]) + float64(a[i+1])*float64(b[i+1]) + float64(a[i+2])*float64(b[i+2]) + float64(a[i+3])*float64(b[i+3])
     i += 4
   while i < len:
-    sum += a[i] * b[i]
+    sum += float64(a[i]) * float64(b[i])
     inc i
-  return -float64(sum)  # negative because we want to minimize
+  return -sum  # negative because we want to minimize
 
 proc manhattanDistance*(a, b: Vector): float64 =
-  var sum: float32
+  var sum: float64
   let len = min(a.len, b.len)
   var i = 0
   while i + 3 < len:
-    sum += abs(a[i]-b[i]) + abs(a[i+1]-b[i+1]) + abs(a[i+2]-b[i+2]) + abs(a[i+3]-b[i+3])
+    sum += abs(float64(a[i])-float64(b[i])) + abs(float64(a[i+1])-float64(b[i+1])) + abs(float64(a[i+2])-float64(b[i+2])) + abs(float64(a[i+3])-float64(b[i+3]))
     i += 4
   while i < len:
-    sum += abs(a[i] - b[i])
+    sum += abs(float64(a[i]) - float64(b[i]))
     inc i
   return sum