Collaboration diagram for duckdb::DistinctStatistics:

[legend]

Public Member Functions
	DistinctStatistics (unique_ptr< HyperLogLog > log, idx_t sample_count, idx_t total_count)

void	Merge (const DistinctStatistics &other)

unique_ptr< DistinctStatistics >	Copy () const

void	UpdateSample (Vector &new_data, idx_t count, Vector &hashes)

void	Update (Vector &new_data, idx_t count, Vector &hashes)

string	ToString () const

idx_t	GetCount () const

void	Serialize (Serializer &serializer) const

Static Public Member Functions
static bool	TypeIsSupported (const LogicalType &type)

static unique_ptr< DistinctStatistics >	Deserialize (Deserializer &deserializer)

Public Attributes
unique_ptr< HyperLogLog >	log
	The HLL of the table.

atomic< idx_t >	sample_count
	How many values have been sampled into the HLL.

atomic< idx_t >	total_count
	How many values have been inserted (before sampling)

Private Member Functions
void	UpdateInternal (Vector &update, idx_t count, Vector &hashes)

Static Private Attributes
static constexpr double	BASE_SAMPLE_RATE = 0.1
	For distinct statistics we sample the input to speed up insertions.

static constexpr double	INTEGRAL_SAMPLE_RATE = 0.3
	For integers, we sample more: likely to be join keys (and hashing is cheaper than, e.g., strings)

The documentation for this class was generated from the following file:

Public Member Functions

Static Public Member Functions