duckdb 1.1.2

v1.1.2 Bugfix Release

on Python PyPI

This is a bug fix release for various issues discovered after we released 1.1.1. There are no new major features, just bug fixes. Database files created by DuckDB versions all the way back to v0.9.* can be read by DuckDB v1.1.2.

What's Changed

• [CI] Re-enable ART zero initialisation verification by @taniabogatsch in #14031
• Push filters instead of overwriting filters by @Tmonster in #14078
• Fix test by @Tmonster in #14079
• fix maximum_threads test inside containers by @wenjun93 in #14083
• Fix: remove is_probe_in_domain by @Light-City in #14084
• Add duckdb_extension.h to amalgamation release by @samansmink in #14086
• Bump minimum required cmake version by @abramk in #14089
• Fix parser error by removing alias by @taniabogatsch in #14090
• [Dev] Move EnumTypeInfoTemplated definition into a hpp file by @Tishj in #14038
• Fix #14077: correctly reset next pointer when reconstructing new row group segment tree after vacuum by @Mytherin in #14092
• Format CSV error messages by @szarnyasg in #14097
• Fix the answer file for tpcds q67 at sf100 by @Tmonster in #14096
• Add v1.1.1 to version_map.json by @Flogex in #14110
• CREATE TABLE now supports columns with ENUM[] types. by @Tishj in #14102
• fix parquet cardinality when first file is empty by @wenjun93 in #14058
• [Python Dev] Make sure the GIL is released when the connection+db are being shut down by @Tishj in #14113
• Less salt by @lnkuiper in #14173
• remove redundant code by @flashmouse in #14172
• comparison of nested types returns true or false always (even with nulls) by @Tmonster in #14094
• subtype DBInterface.Connection in julia client by @piever in #14193
• [Python] Fix a bug with python_scan_all_frames reaching the bottom of the frame stack by @Tishj in #14183
• [Dev] Fix issue where the InsertStatement::ToString call destroyed the alias of the ValuesList by @Tishj in #14171
• [Python] Fix issue related to scanning float64 dtype columns that contain a mask by @Tishj in #14170
• Fix some warnings found while compiling duckdb-node by @carlopi in #13994
• fix minmax type info miss by @Damon07 in #14159
• fix: ArrowSchemaMetadata::GetOption to return empty string instead of raising exception if key is not found. by @rustyconover in #14157
• Issue 14151: Fix conflicting defines on Windows hidden by cmake unity builds by @zmajeed in #14154
• Issue 14189: Fix build when threads are disabled by @zmajeed in #14190
• Fix an uncaught error with a generated column containing a subquery by @Tishj in #14198
• Add missing word in TableFunction comment by @JelteF in #14210
• add method to check whether julia connection is open by @piever in #14195
• Avoid schema changes with IF NOT EXISTS by @ywelsch in #14143
• Fix typos in code by @c8ef in #14243
• [Dev] Add the ExecutorException class, making use of the EXECUTOR ExceptionType by @Tishj in #14231
• [Python] Don't allow construction of DuckDBPyType from empty Dict type by @Tishj in #14221
• Fix #14232: fix deliminator optimizer by @flashmouse in #14238
• [CSV Reader] Also use figure-out-line code when ignoring errors. by @pdet in #14184
• remove redundant Bit::SetBit by @flashmouse in #14226
• Fix #14212: mention correct query component when using literal in DISTINCT ON by @Mytherin in #14255
• Removing overzealous check in Parquet by @hannes in #14268
• Update sqlsmith extension and patches by @Tmonster in #14270
• Support for duckdb.varint extension in Arrow. by @pdet in #14273
• Disable CSV ignore_errors benchmark by @pdet in #14277
• Only slice initialized vectors in PhysicalHashAggregate::SinkDistinctGrouping by @lnkuiper in #14289
• Fix #14249: return NAN when dividened is 0 by @flashmouse in #14298
• Revert "Fix #14249: return NAN when dividened is 0" by @Mytherin in #14308
• fix macro name with the same function name in it which causing repeat… by @Damon07 in #14296
• Fixing issue with the sniffer on copy statetements by @pdet in #14295
• Json bugfixes by @lnkuiper in #14288
• [Bitstring] Add overload for bitstring to accept BIT as the type of the first argument by @Tishj in #14247
• [Fix] Don't initialize reference, constant, and parameter children in intermediate chunk by @taniabogatsch in #14254
• Profiling - correct settings per node type and minor renaming for clarity by @taniabogatsch in #14290
• Fix extension size increase by @samansmink in #14185
• Add option to ignore GeoParquet, disable spatial autoloading when reading GeoParquet by @Maxxen in #14297
• More defensive programming in RowVersionManager::CleanupAppend by @Mytherin in #14317
• Change Makefile to correctly handle DISABLE_SANITIZER and DISABLE_UBSAN by @JelteF in #14316
• [CSV Reader] Making escape not limited to only quotes by @pdet in #14314
• allow external cardinality information (e.g. from iceberg) by @peterboncz in #14292
• [SecretManager] Fix deserialization of Value types in KeyValueSecret::Deserialize by @Tishj in #14332
• Avoid throwing InternalException on reading secret by @carlopi in #14336
• delay the rewrite of a large IN-clause into a MarkJoin on remote Filter-Scans by @peterboncz in #14266
• JSON reader - never generate maps if map_inference_threshold is -1 by @Mytherin in #14348
• [Appender] Support appending to table with generated columns by @Tishj in #14346
• Internal #3251: DateDiff Across Epoch by @hawkfish in #14338
• Bump azure and delta extensions commits by @carlopi in #14350
• Bump spatial to 3f94d52aa9f7d67b1a30e6cea642bbb790c04aa2 by @carlopi in #14351
• Bump more extensions: iceberg, vss and sqlite_scanner by @carlopi in #14352
• Emit profiling info for extension operators by @Mytherin in #14355

Full Changelog: v1.1.1...v1.1.2