remove unused code

src/execution/operator/helper/physical_reservoir_sample.cpp

@@ -1,7 +1,5 @@
 #include "duckdb/execution/operator/helper/physical_reservoir_sample.hpp"
 #include "duckdb/execution/reservoir_sample.hpp"
-#include "duckdb/common/atomic.hpp"
-#include "iostream"
 
 namespace duckdb {
 
@@ -112,25 +110,6 @@ SinkCombineResultType PhysicalReservoirSample::Combine(ExecutionContext &context
 	return SinkCombineResultType::FINISHED;
 }
 
-//static void PrintSampleCount(vector<unique_ptr<BlockingSample>> &samples) {
-//	idx_t total_count = 0;
-//	for (auto &sample : samples) {
-//		total_count += sample->get_sample_count();
-//	}
-//	std::cout << "samples in reservoir is " << total_count << std::endl;
-//}
-
-//static void PrintAllSeenSamples(ReservoirSamplePercentage &sample_percentage) {
-//	idx_t finished_samples_seen = 0;
-//	for (auto &sample : sample_percentage.finished_samples) {
-//		//		auto &tmp = sample->Cast<ReservoirSamplePercentage>();
-//		finished_samples_seen += sample->base_reservoir_sample.num_entries_seen_total;
-//		D_ASSERT(sample->base_reservoir_sample.num_entries_seen_total == 100000);
-//	}
-//	finished_samples_seen += sample_percentage.current_count;
-//	std::cout << "this sample has seen " << finished_samples_seen << " tuples" << std::endl;
-//}
-
 SinkFinalizeType PhysicalReservoirSample::Finalize(Pipeline &pipeline, Event &event, ClientContext &context,
                                                    OperatorSinkFinalizeInput &input) const {
 	auto &global_state = input.global_state.Cast<SampleGlobalSinkState>();

src/execution/reservoir_sample.cpp

@@ -1,6 +1,5 @@
 #include "duckdb/execution/reservoir_sample.hpp"
 #include "duckdb/common/pair.hpp"
-#include "iostream"
 
 namespace duckdb {
 
@@ -47,17 +46,6 @@ void ReservoirSample::AddToReservoir(DataChunk &input) {
 	}
 }
 
-void ReservoirSample::MergeUnfinishedSamples(unique_ptr<BlockingSample> &other) {
-	auto &reservoir_sample = other->Cast<ReservoirSample>();
-	reservoir_sample.Finalize();
-	throw InternalException("a normal reservoir sample should not merge unfisihed samples");
-	//	auto chunk = reservoir_sample.GetChunk();
-	//	while (chunk) {
-	//		AddToReservoir(*chunk);
-	//		chunk = other->GetChunk();
-	//	}
-}
-
 void ReservoirSample::InitializeReservoirWeights() {
 	base_reservoir_sample.InitializeReservoir(num_added_samples, num_added_samples);
 }
@@ -256,31 +244,6 @@ ReservoirSamplePercentage::ReservoirSamplePercentage(Allocator &allocator, doubl
 	current_sample = make_uniq<ReservoirSample>(allocator, reservoir_sample_size, random.NextRandomInteger());
 }
 
-void ReservoirSamplePercentage::MergeUnfinishedSamples(unique_ptr<BlockingSample> &other) {
-	auto &percentage_sample = other->Cast<ReservoirSamplePercentage>();
-	auto other_seen = percentage_sample.current_count;
-	auto other_added = percentage_sample.current_sample->num_added_samples;
-	auto difference = other_seen - other_added;
-	auto chunk = other->GetChunk();
-	while (chunk) {
-		AddToReservoir(*chunk);
-		chunk = other->GetChunk();
-	}
-	current_count += difference;
-}
-
-void ReservoirSamplePercentage::Merge(unique_ptr<BlockingSample> &other) {
-	//! We are now merging all the samples. 80% of every sample should equal 80%
-	//! of all rows so we set sample percentage to 1, which will means every tuple
-	//! in the added chunks will be added
-	sample_percentage = 1;
-	auto chunk = other->GetChunk();
-	while (chunk) {
-		AddToReservoir(*chunk);
-		chunk = other->GetChunk();
-	}
-}
-
 void ReservoirSamplePercentage::AddToReservoir(DataChunk &input) {
 	base_reservoir_sample.num_entries_seen_total += input.size();
 	if (current_count + input.size() > RESERVOIR_THRESHOLD) {
@@ -399,9 +362,6 @@ void BaseReservoirSampling::InitializeReservoir(idx_t cur_size, idx_t sample_siz
 		}
 		SetNextEntry();
 	}
-	if (cur_size > sample_size) {
-		throw InternalException("cur_size should eventually be equal to sample size right?");
-	}
 }
 
 void BaseReservoirSampling::SetNextEntry() {

src/include/duckdb/execution/reservoir_sample.hpp

@@ -75,8 +75,6 @@ class BlockingSample {
 
 	virtual void Merge(unique_ptr<BlockingSample> &other) = 0;
 
-	virtual void MergeUnfinishedSamples(unique_ptr<BlockingSample> &other) = 0;
-
 	virtual void InitializeReservoirWeights() = 0;
 
 	virtual idx_t get_sample_count() = 0;
@@ -114,9 +112,6 @@ class ReservoirSample : public BlockingSample {
 	//! When collecting samples in parallel, merge samples to create a final sample for the column
 	void Merge(unique_ptr<BlockingSample> &other) override;
 
-	//! When merging sample, unfinished sample merging requires special merging logic
-	virtual void MergeUnfinishedSamples(unique_ptr<BlockingSample> &other) override;
-
 	//! Fetches a chunk from the sample. Note that this method is destructive and should only be used after the
 	//! sample is completely built.
 	unique_ptr<DataChunk> GetChunk() override;
@@ -161,10 +156,8 @@ class ReservoirSamplePercentage : public BlockingSample {
 	void AddToReservoir(DataChunk &input) override;
 
 	//! When collecting samples in parallel, merge samples to create a final sample for the column
-	void Merge(unique_ptr<BlockingSample> &other) override;
-
-	//! When merging sample, unfinished sample merging requires special merging logic
-	virtual void MergeUnfinishedSamples(unique_ptr<BlockingSample> &other) override;
+	//! Merge should not be called on reservoir sample percentages. One sample is kept in global state.
+	void Merge(unique_ptr<BlockingSample> &other) override {};
 
 	//! Fetches a chunk from the sample. Note that this method is destructive and should only be used after the
 	//! sample is completely built.
@@ -191,7 +184,6 @@ class ReservoirSamplePercentage : public BlockingSample {
 	//! The fixed sample size of the sub-reservoirs
 	idx_t reservoir_sample_size;
 
-public:
 	//! The current sample
 	unique_ptr<ReservoirSample> current_sample;
 
@@ -202,7 +194,6 @@ class ReservoirSamplePercentage : public BlockingSample {
 	//! Whether or not the stream is finalized. The stream is automatically finalized on the first call to GetChunk();
 	bool is_finalized;
 
-	void PrintSampleCount();
 };
 
 } // namespace duckdb

test/sql/sample/reservoir_testing_rows_value.test

@@ -11,7 +11,7 @@ statement ok
 CREATE or replace TABLE t1 as select range a from range(1000);
 
 query I
-SELECT count(*) from t1 using sample 00;
+SELECT count(*) from t1 using sample 0;
 ----
 0
 
@@ -83,4 +83,4 @@ select count(*) from t1 using sample 700000;
 ----
 700000
 
-endloop
+endloop