You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
support Dictionary types in approx_distinct #24730
Is your feature request related to a problem or challenge?
approx_distinct returns a not implemented error when the input column has a Dictionary type (e.g. Dictionary(Int32, Utf8)), which is common for low-cardinality string columns stored in Parquet or produced by joins/filters. Today SELECT approx_distinct(dict_col) FROM t; Error: Support for 'approx_distinct' for data type Dictionary(Int32, Utf8) is not implemented
Describe the solution you'd like
approx_distinct uses create_hashes internally (via HLLAccumulator and HllGroupsAccumulator), which already handles Dictionary arrays correctly by hashing the logical decoded values. The fix is to route Dictionary(_, _) to HLLAccumulator
Activity
Rich-T-kid commentedon Aug 27, 2026
closing in favor of #22989