changed how initialization works for sketching

Author: Gillgamesh

include/sketch/sketch_columns.h

@@ -46,6 +46,10 @@ class FixedSizeSketchColumn {
     //no-op
   };
 
+  inline bool is_initialized() const {
+    return buckets != nullptr;
+  }
+
   [[deprecated]]
   void zero_contents() {
     clear();
@@ -123,6 +127,10 @@ class ResizeableSketchColumn {
     return os;
   }
 
+  inline bool is_initialized() const {
+    return buckets != nullptr;
+  }
+  
   bool operator==(const ResizeableSketchColumn &other) const {
     size_t other_depth = other.get_depth();
     if (get_depth() != other_depth) {
@@ -148,7 +156,7 @@ class ResizeableAlignedSketchColumn {
   uint8_t capacity;
 public:
   void set_seed(uint64_t new_seed) { seed = new_seed; };
-  const uint64_t get_seed() { return seed; };
+  uint64_t get_seed() const { return seed; };
 
   ResizeableAlignedSketchColumn(uint8_t start_capacity, uint64_t seed);
   ResizeableAlignedSketchColumn(const ResizeableAlignedSketchColumn &other);
@@ -166,6 +174,10 @@ class ResizeableAlignedSketchColumn {
     clear();
   }
 
+  inline bool is_initialized() const {
+    return aligned_buckets != nullptr;
+  }
+
   void reset_sample_state() {
     //no-op
   };

include/sketch/sketch_concept.h

@@ -40,6 +40,8 @@ concept SketchColumnConcept = requires(T t, T other) {
   { t.sample() } -> std::same_as<SketchSample<V>>;
   { t.update(std::declval<V>()) } -> std::same_as<void>;
   { t.merge(other) } -> std::same_as<void>;
+  
+  { t.is_initialized() } -> std::same_as<bool>;
 
   { t.clear()} -> std::same_as<void>;
   { t.zero_contents()} -> std::same_as<void>;

src/sketch_columns.cpp

@@ -2,19 +2,31 @@
 
 FixedSizeSketchColumn::FixedSizeSketchColumn(uint8_t capacity, uint64_t seed) :
     capacity(capacity), seed(seed) {
-  buckets = new Bucket[capacity];
-  std::memset(buckets, 0, capacity * sizeof(Bucket));
+  likely_if (capacity > 0) {
+    buckets = new Bucket[capacity];
+    std::memset(buckets, 0, capacity * sizeof(Bucket));
+  }
+  else {
+    buckets = nullptr;
+  }
 }
 
 FixedSizeSketchColumn::FixedSizeSketchColumn(const FixedSizeSketchColumn &other) :
     capacity(other.capacity), seed(other.seed), deterministic_bucket(other.deterministic_bucket) {
-  buckets = new Bucket[capacity];
-  std::memcpy(buckets, other.buckets, capacity * sizeof(Bucket));
+  likely_if (capacity > 0) {
+    buckets = new Bucket[capacity];
+    std::memcpy(buckets, other.buckets, capacity * sizeof(Bucket));
+  }
+  else {
+    buckets = nullptr;
+  }
 }
 
 FixedSizeSketchColumn::FixedSizeSketchColumn(FixedSizeSketchColumn &&other) :
     capacity(other.capacity), seed(other.seed), deterministic_bucket(other.deterministic_bucket) {
-    buckets = std::move(other.buckets);
+      buckets = other.buckets;
+      other.buckets = nullptr;
+      other.capacity = 0;
 }
 
 FixedSizeSketchColumn& FixedSizeSketchColumn::operator=(FixedSizeSketchColumn &&other) {
@@ -26,6 +38,7 @@ FixedSizeSketchColumn& FixedSizeSketchColumn::operator=(FixedSizeSketchColumn &&
 
     buckets = other.buckets;
     other.buckets = nullptr;
+    other.capacity = 0;
   }
   return *this;
 }
@@ -84,24 +97,35 @@ void FixedSizeSketchColumn::update(const vec_t update) {
   deterministic_bucket ^= {update, checksum};
 }
 
-
-ResizeableSketchColumn::ResizeableSketchColumn(uint8_t start_capacity, uint64_t seed) :
-    capacity(start_capacity), seed(seed) {
+ResizeableSketchColumn::ResizeableSketchColumn(uint8_t start_capacity,
+                                               uint64_t seed)
+    : capacity(start_capacity), seed(seed) {
+  likely_if(capacity > 0) {
     buckets = new Bucket[start_capacity];
     std::memset(buckets, 0, capacity * sizeof(Bucket));
+  }
+  else {
+    buckets = nullptr;
+  }
 }
 
 ResizeableSketchColumn::ResizeableSketchColumn(const ResizeableSketchColumn &other) :
     capacity(other.capacity), seed(other.seed), deterministic_bucket(other.deterministic_bucket) {
-  buckets = new Bucket[capacity];
-  std::memcpy(buckets, other.buckets, capacity * sizeof(Bucket));
+  likely_if(capacity > 0) {
+    buckets = new Bucket[capacity];
+    std::memcpy(buckets, other.buckets, capacity * sizeof(Bucket));
+  }
+  else {
+    buckets = nullptr;
+  }
 }
 
 ResizeableSketchColumn::ResizeableSketchColumn(ResizeableSketchColumn &&other) :
     capacity(other.capacity), seed(other.seed), deterministic_bucket(other.deterministic_bucket) {
     // move constructor
     buckets = other.buckets;
     other.buckets = nullptr;
+    other.capacity = 0;
 }
 
 ResizeableSketchColumn& ResizeableSketchColumn::operator=(ResizeableSketchColumn &&other) {
@@ -113,6 +137,7 @@ ResizeableSketchColumn& ResizeableSketchColumn::operator=(ResizeableSketchColumn
 
     buckets = other.buckets;
     other.buckets = nullptr;
+    other.capacity = 0;
   }
   return *this;
 }