Data and AI struggles

Trouble with DuckDB column

Databases9 posts from 4 people6d active+00 posts in the last 7 days, 4 the 7 days before (fading)

Posts per day

Posts per day9 posts, Aug 31 to Oct 21

The posts behind it

9, newest first
GitHub issuesOct 14 7 days agoINTERNAL Error "dict-surviving cache: global dictionary column changed identity" when a LEFT blockwise NL join consumes a dictionary-surviving RIGHT hash join...what `SET disabled_optimizers = 'statistics_propagation'` returns. Physical plan of the failing query: Same data, what changes the outcome: | variation | result | |---|---| | as above | internal error | | `SET threads = 1` / 4 / 8 | internal error | | `d.d_name` as DATE or...duckdb/duckdb · h4nsmullerOpen ↗GitHub issuesOct 14 7 days agoParquet writer dictionary encoding merges INTERVAL values that are only equal after normalisation: a column with both '1 month 2 days' and '32 days' is written back as 2000 x '32 days'### What happens? The Parquet writer's dictionary encoding merges `INTERVAL` values that are *different values* but compare equal under DuckDB's normalised interval comparison (`INTERVAL '1 month 2 days' = INTERVAL '32 days'` is true because a month is normalised to 30 days)....duckdb/duckdb · h4nsmuller1 repliesOpen ↗GitHub issuesOct 11 10 days agoOptimizer: NOT (a = const AND b IN (...)) under a subquery plus a scalar subquery fails with "unordered_map::at" (join_order after in_clause rewrite); long IN lists give INTERNAL Error: Failed to bind column reference "IN (...)"### What happens A filter with `NOT (a = const AND b IN (...))` under a subquery, combined with a scalar subquery in the outer `WHERE`, fails during optimization: Expected 2500. `SET disabled_optimizers = 'join_order'` returns 2500. The same query without the nesting (all...duckdb/duckdb · h4nsmullerOpen ↗GitHub issuesOct 9 12 days ago2.0 alpha: comparing a CASE/split_part result to a constant fails with "BindBuiltinScalarFunction did not return a BoundFunctionExpression" when the column is NULL in every row...compares a string derived from a column to a constant fails with an internal-looking error when that column happens to be NULL in every row of the table: The query is valid SQL and returns the expected result on 1.5.6. It also works on 2.0 as soon as a single row has a non-NULL...duckdb/duckdb · dxts2 repliesOpen ↗GitHub issuesOct 2 2 weeks agoManualVuln33 Segmentation Fault in PIVOT When Using an Empty ENUM Type as the Pivot Column### What happens? `Binder::BindPivot` in `src/planner/binder/tableref/bind_pivot.cpp` calls `EnumType::GetSize(type)` (line 652) to determine how many pivot columns to generate. When the ENUM type is empty (size 0), the entry-building loop on lines 653-659 never executes,...duckdb/duckdb · ylwango613Open ↗GitHub issuesOct 2 2 weeks agoManualVuln19 CSV Scanner INVALID_ENCODING Handler Uses File Column Index to Access Projected-Column-Sized file_types Vector### What happens? `LineError::HandleErrors()` (`src/execution/operator/csv_scanner/scanner/string_value_scanner.cpp` around line 750) handles `INVALID_ENCODING` errors with: `col_idx` is set from `cur_error.col_idx`, which is the **file column index** (e.g. 28 for column C28 in...duckdb/duckdb · ylwango613Open ↗GitHub issuesOct 1 2 weeks agoManualVuln12 INTERNAL Error in CREATE TABLE When COMPRESSION_COUNT Sentinel Is Used as Column Compression Type...it to functions that do not handle it. The parser transformer and the binder both fail to do this, allowing user-controlled SQL to trigger an INTERNAL error. ### To Reproduce Reproducible directly via plain SQL from the DuckDB CLI client, no external file or script required. SQL...duckdb/duckdb · ylwango613Open ↗GitHub issuesAug 31 7 weeks agoUnexpected results when comparing a LIST column### What happens? *Found by an automated fuzzing/triage agent; the repro is verified, but the analysis may be wrong.* In a single LIST column, two identical values compared against the same constant return different answers. With rows `[]`, `[-2, -1]`, `[-2, -1]`, the predicate...duckdb/duckdb · suyZhongOpen ↗GitHub issuesAug 31 7 weeks agoUnexpected results when a sampled scan of an indexed column filtered by `IN (VALUES ...)`### What happens? On a 16384-row table with an index on `c1`, `TABLESAMPLE SYSTEM (10%) REPEATABLE (7)` returns 0 rows, and `WHERE c1 = true` over that sample also returns 0. The logically equivalent `WHERE c1 IN (VALUES (true))` returns 2048 - the count of matching rows in the...duckdb/duckdb · suyZhongOpen ↗

Companies and products named

About this problem

Evidence

9 posts from 4 people in 1 place, about 1 a week over 45 days. Mostly on duckdb/duckdb. Tools named alongside: AWS.

Frustration Frustration 0 of 3· Seen on GitHub issues

How it was grouped

Posts about DuckDB with no known issue, sharing the word "column". First post Aug 31, 2026, latest Oct 14, 2026. Corroborated: 3 or more posts from 2 or more people or places. Method

History

MomentumOct 21, 2026steady to fading
MomentumOct 18, 2026rising to steady
MomentumOct 15, 2026fading to rising
MomentumOct 9, 2026rising to fading
AddedOct 7, 2026Trouble with DuckDB column (3 posts)