---
name: ClickHouse Changelog
slug: clickhouse-changelog
type: scrape
source_url: https://clickhouse.com/docs/whats-new/changelog
organization: ClickHouse
organization_slug: clickhouse
total_releases: 10
latest_version: 26.10
latest_date: 2026-10-22
last_updated: 2026-11-06
tracking_since: 2026-03-26
canonical: https://releases.sh/clickhouse/clickhouse-changelog
organization_url: https://releases.sh/clickhouse
---

<Summary type="rolling" window-days="90" release-count="1">
ClickHouse 26.3 LTS tightened data serialization and SQL semantics. String serialization now propagates into nested types like Array and Map, making downgrades unsafe but upgrades transparent. The release also corrected NOT operator precedence to match SQL standard—queries relying on the old non-standard binding will produce different results—and removed the experimental `hypothesis` skip index type and `detectProgrammingLanguage` function.
</Summary>

<Summary type="monthly" period="March 2026" release-count="1">
ClickHouse 26.3 LTS tightened data type handling and SQL compliance. String serialization versions now propagate to nested types within Arrays, Maps, and Variants—upgrades are safe but downgrades will break access to 26.3-written data. The release also fixed NOT operator precedence to match SQL standards, removed the hypothesis skip index type, and eliminated the experimental detectProgrammingLanguage function.
</Summary>

<Release version="26.10" date="October 22, 2026" published="2026-10-22T00:00:00.000Z" url="https://clickhouse.com/docs/whats-new/changelog#26-10">
## ClickHouse release 26.10

## Backward Incompatible Changes

- Query analysis preceding the analyzer is removed; `enable_analyzer` must be `1` on every server before upgrading a mixed cluster. `use_analyzer_for_mutations` is removed and `parallel_replicas_only_with_analyzer` is obsolete.
- Predicate push-down on the AST for non-SELECT queries removed; `enable_optimize_predicate_expression` is obsolete.
- `system.metric_log` now uses the `bucketed` schema by default (set `<schema_type>wide</schema_type>` to keep the old schema).
- `seccomp` system call filter applied at startup; packaged config sets it to `trap`.
- Settings constraints and `allow_feature_tier` enforced for SQL-defined users, roles and profiles.
- Named collection key overrides require `SHOW NAMED COLLECTIONS SECRETS`; `allow_named_collection_override_by_default` is obsolete.
- `max_bytes_before_external_join` now applies to every hash-based join algorithm including `grace_hash`.
- `arrayCount` returns `UInt64` instead of `UInt32`.
- `accurateCast` from String to Nullable now throws instead of returning NULL.
- `SELECT ... FINAL` on `SummingMergeTree` no longer removes all-zero rows.
- `skip_unavailable_shards` now fails with `ALL_CONNECTION_TRIES_FAILED` when all shards are skipped.
- MATERIALIZED columns depending on TTL-reset columns are recomputed.
- JSON null into non-Nullable numeric column raises with `input_format_null_as_default = 0`.
- `ALTER ... ADD/MODIFY COLUMN` honors `data_type_default_nullable`.
- `arrayJoin` lowered to ARRAY JOIN step by default.
- Row policies on Distributed tables read through Merge fail with `ILLEGAL_PREWHERE`.
- `clickhouse-local --ignore-error` now reports errors; `clickhouse-client --ignore-error` exit code changes.
- `ATTACH TABLE ... FROM` requires `READ/WRITE ON FILE`.
- MySQL dictionaries cannot enable `enable_local_infile` via CREATE DICTIONARY.
- Disk config elements are validated (`UNKNOWN_ELEMENT_IN_CONFIG`).
- Many settings made obsolete or removed (`validate_mutation_query`, `query_plan_read_in_order`, `use_compact_format_in_distributed_parts_names`, `s3_max_connections`, etc.); `year()` zero-argument, `system.codecs.is_experimental`, and `detectTonality` removed.
- `ZXC` codec format changed; previously written data cannot be read.
- MySQL SSL CA and hostname verification now enforced.

## New Features

- Sets for `IN` spill to disk by default past half of memory limit.
- Trino SQL dialect parts supported (`TRY_CAST`, `ARRAY[...]`, aliases).
- Block nested loop join for arbitrary `ON` conditions (`allow_block_nested_loop_join`).
- Function composition operator `f | g` and argument placeholders.
- Query cache can store entries on disk.
- Notebook mode in the Web UI (`/play`).
- SQLite format support; `Cluster` and `Overlay` database engines.
- Workload query-aware scheduling and spilling per workload.
- Quota `written_rows` resource.
- Temporary table limits settings.
- `SIMILAR TO` predicate.
- `chdig` v26.10.1 integrated.
- Parquet VARIANT and FILE logical type support; Iceberg variant type.
- `generateRandom` supports JSON, Dynamic, Variant, BFloat16, Time types.
- `obfuscate` table function; `mapRemove`, `mapEntries`, `arrayMinIndex`, `arrayMaxIndex`, `arrayRemoveAt`, pack-bits functions.
- Many more functions, PromQL and system table additions.

## Experimental Features

- Columns cache, `LATERAL JOIN`, `logsql` dialect, `PCO` codec, unity catalog v2, embedded Iceberg REST catalog, XGBoost dictionaries, IPCrypt functions, Iceberg deletion vectors and DROP PARTITION, distributed plan improvements, and PromQL additions.

## Performance Improvements

- Hash tables for GROUP BY/JOIN use up to 2x less memory; many GROUP BY, JOIN, and aggregation speedups; `apply_string_filters_during_scan` enabled by default; `query_plan_join_shard_by_pk_ranges` enabled by default; text-index postings cache enabled by default.
</Release>

<Release version="26.9" date="September 21, 2026" published="2026-09-21T00:00:00.000Z" url="https://clickhouse.com/docs/whats-new/changelog#26-9">
## ClickHouse release 26.9

## Backward Incompatible Changes
- The analyzer can no longer be disabled; `enable_analyzer` / `allow_experimental_analyzer` are obsolete.
- Default compression changed from `LZ4` to `ZSTD(3)` (size-aware for MergeTree column data: LZ4 below 100 MB, ZSTD(3) at or above).
- Removed the CatBoost integration (`catboostEvaluate`, `system.models`, `SYSTEM RELOAD MODEL(S)`). Revoke grants and drop dependent objects before upgrading.
- Removed the `read_resource`/`write_resource` disk options; use `CREATE RESOURCE`.
- Removed the WasmEdge engine; `webassembly_udf_engine = 'wasmedge'` prevents startup.
- Removed experimental `WINDOW VIEW`, `WATCH`, and `windowID`; drop window views before upgrading.
- Removed sharded GROUP BY aggregation (`enable_sharding_aggregator` is now a no-op).
- `runningConcurrency` is now non-deterministic; it is rejected in sorting/partition keys.
- `BACKUP`/`RESTORE` to `Disk(...)` now require `SOURCES` grants.
- Compression codecs gated by per-codec `enable_<codec>_codec` settings; `allow_experimental_codecs` is obsolete.
- `rename_files_after_processing` requires `WRITE ON FILE`.
- `system.query_log` / `query_thread_log` / `processes` `interface` and `http_method` columns changed to `Enum8`.
- `CREATE TABLE ... AS mergeTreeIndex(...)` and `AS timeSeries*(...)` forms are no longer accepted.
- Removed `s3_disable_checksum`.
- `Nullable(Tuple(...))` is GA and enabled by default.
- New `validate_group_by_all_key_types` setting.

## New Features
- `DISTINCT` can spill to disk.
- Refreshable materialized views support `REFRESH ... APPEND INCREMENTAL`.
- New `LIMIT ... AFTER` / `UNTIL` forms.
- `CREATE TOKEN` statement and `GRANTS` clause for authentication methods.
- Regression aggregate functions `regr_*`.
- MergeTree `max_table_size_*` settings and database `max_tables` setting.
- JSON bracket subcolumn syntax; `system.statements` table.
- `skip_empty_columns_on_insert` MergeTree setting; `arrayFlattenedLength`; `parseISO8601Duration`.
- `clickhouse-client --ssh-key-file` without a value.
- `keyValuePairs` text index tokenizer.
- BACKUP/RESTORE for workloads and resources; `workload_admission_timeout_ms`.
- S3Queue `exclusive` mode; NATS TLS/client cert settings.

## Experimental
- Prometheus API endpoints for TimeSeries; `/ui` SQL console; reimplemented KQL dialect; Trino dialect.
- `hierarchicalKMeans`, `assignCentroid`; SELECT on TimeSeries; more PromQL functions.
- Distributed plan improvements; `Merge` tables with parallel replicas; silk fiber runtime.
- Delta Lake CREATE TABLE support.

## Performance
- Many improvements to aggregations, joins, text indexes, JSON, Parquet, Iceberg, and string/date functions.
</Release>

<Release version="26.6" date="June 25, 2026" published="2026-06-25T00:00:00.000Z" url="https://clickhouse.com/docs/whats-new/changelog#26-6">
## ClickHouse release 26.6

#### Backward Incompatible Changes

* Removed `allow_experimental_query_deduplication` feature after long experimental period with no tests or support.
* XRay-based `SYSTEM INSTRUMENT` feature renamed from `parameters` to `arguments`: `system.instrumentation.parameters` → `arguments`, and `SYSTEM INSTRUMENT ADD ... HANDLER PARAMETERS ...` → `SYSTEM INSTRUMENT ADD ... HANDLER ARGUMENTS ...`
* Setting `show_data_lake_catalogs_in_system_tables` renamed to `show_remote_databases_in_system_tables` with expanded scope: when set to 0 (default), `MySQL` and `PostgreSQL` databases are also hidden from system tables (old name kept as alias).
* Default x86 build now targets x86-64-v3 (AVX2) instead of x86-64-v2 (SSE4.2); requires CPU with AVX2 support. Use `amd64compat` build for older CPUs.
* Nested `Dynamic`/`Variant` now rejected in `min`/`max` aggregates and `minmax` indexes.
* `icebergHash` and `icebergBucket` now explicitly reject `Int128`, `UInt128`, `Int256`, `UInt256`, and `Decimal256` arguments.
* `--echo` CLI option now accepts optional boolean value; positional queries after bare `--echo` treated as its value; use `--echo --query "..."` or `--echo=false` instead.
* Removed experimental KQL functions `array_sort_asc`, `array_sort_desc`, and their SQL backends.
* `ALTER TABLE ... REPLACE PARTITION ... FROM ...` now rejects requests when source table has no parts in requested partition (previously silently dropped destination data). Set `allow_replace_partition_from_empty_source = 1` to restore old behavior.
* Default `insert_deduplication_version` changed from `compatible_double_hashes` to `new_unified_hash`; deduplication now per-insert instead of per-part/partition.

#### New Features

* Continuous queries over `MergeTree` tables via snapshot reads (first step toward streaming query support).
* Hypothetical (what-if) skip indexes with `CREATE HYPOTHETICAL INDEX` and `EXPLAIN WHATIF` for cost estimation.
* Embedded `/schema` web UI visualizing dependency graph between tables, views, materialized views, dictionaries, and distributed tables.
* `PNG` output format support for rendering query results as images.
* AI coding agent detection (Claude Code, Cursor, Codex, etc.) reported in `system.query_log`, `system.query_thread_log`, and `system.processes`.
* New `system.documentation` table with embedded reference documentation for functions, table engines, data types, settings, and formats.
* `GeoJSON` input format for reading FeatureCollection documents with native `Geometry` type support.
* Mapbox Vector Tile functions: `MVTEncodeGeom`, `MVTEncode`, `MVTBoundingBox`, `MVTBoundingBoxMercator` (with PostGIS aliases).
* `clickhouse-local` now supports `SYSTEM START LISTEN` and `SYSTEM STOP LISTEN` queries.
* Interactive `help` command in `clickhouse-client` and `clickhouse-local` with syntax-highlighted documentation.
* Azure Data Lake Storage Gen2 write support.
* `RowBinaryWithNamesAndTypesAndDefaults` format for better schema evolution support.
* `ADD ENUM VALUES` in `ALTER TABLE` to append new enum values without respecifying all current values.
* Refreshable materialized view `REFRESH DEPENDS ON` for dependency-triggered refreshes.
* Column selection by name pattern: `* LIKE '<pattern>'` and `* ILIKE '<pattern>'`.
* `ESCAPE` syntax for custom escape characters in `LIKE` patterns.
* `materialize_projections_on_insert` and `materialize_projections_on_merge` MergeTree settings.
* `allow_tuple_element_aggregation` setting for recursive tuple flattening in aggregating engines.
* `quantizeBFloat16ToInt8` and `dequantizeInt8ToBFloat16` scalar codec functions.
* `arrayTopK` and `arrayBottomK` array functions.
* `system.iceberg_files` table exposing per-file metadata for Iceberg tables.
* `system.constraints` table for CHECK and ASSUME constraints.
* `system.dictionary_layouts`, `system.dictionary_sources`, `system.data_skipping_index_types`, `system.disk_types` tables with embedded documentation.
* `SYSTEM RESTART DISK` command for reloading disk metadata and re-scanning data parts.
* PostgreSQL-style `expr OP SOME(array)` / `expr OP ALL(array)` syntax.
* `hostname_longest_common_prefix` and `hostname_longest_common_suffix` load balancing strategies.
* TLS client certificate information in `system.session_log`.
* Optional `external_id` credential for S3 role-based access.
* `S3Queue` setting `after_processing_move_preserve_path` for preserving source paths during moves.
* `message_queue_disable_insertion` setting to disable insertion from message queue engines.
* `LOCALTIME` and `LOCALTIMESTAMP` SQL-standard functions.
* `date_part()` function as syntactic sugar for `EXTRACT()`.
* PostgreSQL-compatible `EXTRACT(TIMEZONE_HOUR/MINUTE FROM dt)` and interval extraction.
* `min_by` and `max_by` aggregate aliases for `argMin` and `argMax`.
* `REGEXP_SUBSTR` case-insensitive alias of `regexpExtract`.
* `SESSION_USER` case-insensitive alias of `currentUser()`.
* Trailing `NULL` / `NOT NULL` modifier in `ALTER TABLE ... ADD/MODIFY COLUMN`.
* `h3PolygonToCellsWithContainment` for H3 polygon-to-cells conversion with containment modes.
* `IPV4_PREFIX_BITS` and `IPV6_PREFIX_BITS` flags for IP-based keying.
* `formatReadableTimeDelta` now accepts `INTERVAL` expressions other than Month/Year.
* Optional precision argument for `formatReadableSize`, `formatReadableDecimalSize`, `formatReadableQuantity`.
* `output_format_float_precision` setting for controlling decimal digits in float output.
* `output_format_always_write_decimal_point_in_float_and_decimal` setting.
* New HTTP handler configuration elements: `<url_prefix>`, `<full_url_prefix>`, `<url_regexp>`, `<full_url_regexp>`, `<headers_regexp>`.
* `basic` statistics (compact per-column min/max, avg string length, NULL counts).
* TTL nodes in ClickHouse Keeper (opt-in via `create_ttl` feature flag).
* Memory reservation feature for workloads.

#### Experimental Features

* Multi-stage distributed query execution with scatter/broadcast/gather/shuffle exchanges via TCP or shared object storage.
* `dphyp` join reordering algorithm for inner joins with `query_plan_optimize_join_order_max_searched_plans` setting.
* `aiEmbed` function for LLM-based embedding generation.
* Linux OOM canary: sacrificial child process to shed memory pressure before main server.
* Web terminal interface (`/webterminal`) promoted to production feature (enabled by default).
* PromQL HTTP endpoints with database/table URL query parameters.
* PromQL `histogram_quantile` function support in `prometheusQuery` and `prometheusQueryRange`.
* Lazy posting list apply mode for text indexes with cursor-based decoding.
* AssemblyScript ABI support for WebAssembly UDFs.
* Prometheus handlers on main HTTP port with optional prefix.
* `packed_skip_index_max_bytes` MergeTree setting for bundling skip-index substreams.
* `Buffers` serialization for WebAssembly UDFs using `ABI BUFFERED_V1`.
* `use_reader_executor` setting routing reads through new pipeline-based `ReaderExecutor`.
* Descending sort order in MergeTree sorting keys (no longer requires `allow_experimental_reverse_key`).
* `Nullable(Tuple(...))` type (Beta, disabled by default).

#### Performance Improvements

* Keeper ~2x faster overall (better batching, message pipelining, log append pipelining).
* Simple `SELECT` queries ~50% faster (reduced per-query overhead for parsing, analysis, planning).
* Primary key index analysis optimized for long and high-cardinality keys.
* New `ipnsort` and `driftsort` stable sorting implementations for faster `ORDER BY`.
* `GROUP BY` optimization for high cardinality evenly distributed keys via row scattering (`enable_sharding_aggregator`).
* Hash table prefetching for string key `GROUP BY` queries (~8% improvement).
* Reduced peak memory for merging partial two-level aggregation results.
* `Enum` DataType memory optimization (up to 10x footprint reduction).
* Identifier resolution caching to prevent duplicate resolution.
* Query analysis ~50x faster for nested SELECT * over tables with ~1200 columns.
* Query analysis faster for queries with many/deeply nested function calls.
* Replicated access storage login/startup stalls fixed with per-entity cache batching.
* Hash join runtime filter from `FixedHashMap` sharing via `join_runtime_filter_from_fixed_hash_table`.
* Lazy selector/replication index application for JOIN + LIMIT scenarios.
* DP JOIN reordering now allowed with parallel replicas.
* JOIN runtime filter aware join-reorder cost model.
* RIGHT/FULL join non-joined row emission de-virtualized via batching.
* Packed `keys32`/`keys64` methods in `HashJoin` and `Set`.
* Optimal nullable column handling in aggregation.
* `LIMIT BY` query performance improvements via multiple optimizations.
* Sharded aggregation, grace_hash join, parallel window partitioning ~2x faster (hardware CRC32C).
* Array distance functions auto-vectorized on x86-64-v4/v3.
* `{Norm}Distance` and `dotProduct` functions improved.
* `arrayNorm` function performance improved.
* AVX-512 const-left path improvements for distance functions.
* `encrypt`, `decrypt`, `tryDecrypt`, `aes_encrypt_mysql`, `aes_decrypt_mysql` up to 10x faster.
* `encrypt`, `decrypt`, `halfMD5` faster via OpenSSL 3.x provider lookup avoidance.
* `SHA1` SIMD implementation using AVX-512 for parallel hashing.
* AArch64 ASIMD/NEON backend for multi-buffer `MD5`.
* `MATERIALIZE PROJECTION` ~3.4x faster via pre-squashing.
* `MergeTree` INSERT latency reduced via skip index column permutation optimization.
* Parquet `BYTE_STREAM_SPLIT` decoding faster via Arrow SIMD paths.
* `LowCardinality` insertion performance improved for `bloom_filter` indexes.
* Blob copy parallelization in `DiskObjectStorageTransaction`.
* `One` input format avoids file content reading.
* macOS startup ~3x faster (no dynamic symbol export).
* Filesystem cache loading faster via redundant directory open elimination.
* Filesystem cache loading faster via idle thread re-utilization.
* Startup overhead reduced via WasmEdge library update.
* MergeTree primary key and skip indexes prune granules for `ifNull`/`coalesce` wrapped conditions.
* DateTime64 skip index evaluation consumes less CPU.
* Text index multi-token search analysis optimized for rare tokens.
* Text index posting-list segment caching.
* Generic exclusion search performance improved.
* S3 client endpoint/bucket region discovery caching.
* `has()` with constant arrays transformed to faster `in()` implementation.
* ThreadPool LIFO wake strategy reduces memory fragmentation.
* Approximate runtime filter and bloom filter index performance improved.
* `ASOF JOIN` can now use `parallel_hash` algorithm.
* Early termination for empty `INTERSECT`/`EXCEPT` results.
* `bitmapContains` for non-UInt64 groupBitmap optimized.
* Faster discontinuous queries for single-dictionary `LowCardinality` columns.
* `HashJoin` probe fast path for single-row blocks.
* Predicate pushdown for remote table function queries.
* `enable_join_transitive_predicates` enabled by default.
* Text index support for `multiSearchAny`, `multiSearchAnyUTF8`, `multiMatchAny`, improved `match` pattern analysis.
* Filter merging into PREWHERE on second optimizer pass.
* `QueryConditionCache` records individually filtered-out granules.
* Keeper `nuraft_use_bg_thread_for_snapshot_io` enabled by default.
* Keeper snapshot application peak memory usage reduced.
* LLVM/JIT allocations routed to dedicated jemalloc arena.
* Excessive memory reservation for sparse dictionary-encoded `Nullable(String)` Parquet columns fixed.
* Cardinality estimation improved for deterministic single-argument functions in join reordering.
* Asynchronous logging CPU overhead reduced via LIFO queue notifications.
* Duplicate calculation removal during query execution.
* Recursive CTE result processing parallelized.
* Case-insensitive substring search faster via NEON (ARM) and wider AVX2 (x86).
* FPC floating-point codec ARM regression fixed.

#### Improvements

* Dynamic server hard memory limit adjustment based on current memory usage and system MemAvailable.
* C# PostgreSQL client support in PostgreSQL protocol.
* Mutations now use the analyzer.
* Schema evolution with concurrent consistency checks for data lake catalogs.
* Cache disk support for datalake table engines.
* DataLakeCatalog `onelake` type now uses Blob endpoint by default.
* `S3(Azure)Queue` default `persistent_processing_node_ttl_seconds` increased from 1 hour to 6 hours.
* Iceberg metadata cache default size reduced from 1 GB to 128 MB.
* S3Queue streaming task `query_id` propagated to dependent table inserts.
* `system.query_log` `used_storages` populated with storage engine name for DataLakeCatalog queries.
* Iceberg snapshot operation and summary information now accessible.
* S3 multipart upload finalization completed asynchronously via task tracker.
* MergeTree text index default parameter settings: `text_index_dictionary_block_size`, `text_index_dictionary_block_frontcoding_compression`, `text_index_posting_list_block_size`, `text_index_posting_list_codec`.
* Projections can override additional MergeTree settings (compression, part format, serialization).
* `system.merges` new columns: `current_projection`, `current_projection_progress`, `projections_completed`, `projections_remaining`.
* PREWHERE with IN subquery on primary key columns now uses primary key index for granule pruning.
* `h3PolygonToCells` enforces max array size, validates H3 return codes, rejects MultiLineString.
* `system.keeper_snapshots` table with local Keeper snapshot information.
* `system.keeper_changelogs` table with Keeper changelog (Raft log) file information.
* `system.keeper_cluster` table with Raft cluster member information.
* Keeper profile events expanded for watches.
* Keeper `coordination_settings` for NuRaft uncommitted log entry admission limiting and append-entries throttling.
* `enable_compression` setting for mysql table function, MySQL table/database engines, and MySQL dictionary source.
* Query cancellation latency reduced for PostgreSQL wire protocol.
* Query cancellation latency reduced for MySQL wire protocol.
* MySQL handshake `auth_response` length reading fixed for values >= 128.
* `MaterializedPostgreSQL` maps PostgreSQL `numeric(p, 0)` with precision > 76 to `Int256`.
</Release>

<Release version="26.5" date="May 21, 2026" published="2026-05-21T00:00:00.000Z" url="https://clickhouse.com/docs/whats-new/changelog#26-5">
## ClickHouse release 26.5

## ClickHouse release 26.5

### Backward Incompatible Changes

- Changed defaults of `date_time_input_format` and `cast_string_to_date_time_mode` from `basic` to `best_effort`. Set to `basic` or `compatibility` to restore old parsing behavior.
- Tuple element name `null` is now forbidden due to conflicts with Nullable subcolumn names.
- Added settings `dynamic_disk_allow_from_env`, `dynamic_disk_allow_from_zk`, `dynamic_disk_allow_include` that disallow these operations by default in dynamic disks.
- Removed obsolete Arrow-based Parquet reader/writer; native implementation is used instead.
- `SHOW CREATE TABLE t` now prefers temporary tables when both exist; `DESCRIBE TEMPORARY TABLE` syntax is now supported.
- Reduced default `http_max_fields` from 1,000,000 to 1,000 and `http_max_field_name_size` from 128 KB to 4 KB to limit pre-authentication memory usage.
- Added `histograms` nested column to `system.metric_log`; deprecates `system.histogram_metric_log` table.
- `CAST` to `DateTime` or `DateTime64` without explicit timezone now preserves source timezone.
- Removed `kql` table function; use `SET dialect = 'kusto'` instead.
- Window functions `RANK` and `DENSE_RANK` now reject arguments per SQL standard.

### New Features

- Added `max_bytes_ratio_before_external_join` setting (default 0.5) for hash join spill-to-disk threshold as fraction of available memory.
- Added `filesystem` table function to represent directory structure as queryable table.
- Allow passing bare function names to higher-order functions like `arrayMap`, `arrayFilter`.
- Added `send_table_structure_on_insert_with_inline_data` setting and `--inline-insert-data` client option.
- Added `tokenizeQuery` and `highlightQuery` functions for SQL query tokenization and syntax highlighting.
- Added `url_base` setting to resolve relative URLs in `url` table function and `URL` table engine.
- Support negative values in `LIMIT BY` clause to select rows from end of each group.
- Support `tuple` and `array` output in `json_value` function.
- Added `kafka_autodetect_client_rack` parameter to discover Availability Zone and propagate as client rack.
- Added `Read` event type to `system.blob_storage_log` for tracking object storage reads.
- Users can now see ZooKeeper watches via new `system.zookeeper_watches` table.
- Added `Shards` profile event counting shards in distributed queries.
- `WASM` UDFs can be declared `DETERMINISTIC` for constant folding.
- Added `parallel_replicas_prefer_local_replica` setting to control replica selection.
- Added socket buffer size settings: `{disk,storage,http}_connections_{rcvbuf,sndbuf}`.
- Support `CREATE OR REPLACE MATERIALIZED VIEW` with atomic swap semantics.
- Added S3 histogram metrics `s3_read_request_duration_microseconds` and `s3_read_request_bytes`.
- Added `Paimon`, `PaimonS3`, `PaimonAzure`, `PaimonHDFS`, `PaimonLocal` table engines with incremental read support.
- Added Kafka `kafka_map_virtual_columns_on_write` setting to map virtual columns on insert.
- Added `SYSTEM PAUSE VIEW` and `SYSTEM PAUSE VIEWS` for refreshable materialized views.
- Added `regexpPosition` function (PostgreSQL-compatible aliases `regexpInstr`, `regexp_instr`) for regex match position.
- New functions `isPrime` and `isProbablePrime` for primality checks.
- `clear` and `/clear` now clear the terminal in CLI tools.
- Allow `file` table function to accept `Array(String)` of paths.
- Added `deterministic` and `higher_order` columns to `system.functions`.
- Added optional encoding variant parameter to `bech32Encode` and raw decode mode to `bech32Decode`.
- Can now write data in `AvroConfluent` format with `output_format_avro_confluent_subject` setting.
- Added `prettyPrintJSON` function to format JSON strings into human-readable form.
- Added `STRING_AGG` as case-insensitive alias of `groupConcat`.
- Individual subquery results can be cached independently with `use_query_cache` per subquery.
- Added lexer-based syntax highlighting to web UI query editor.

### Experimental Features

- Added experimental web terminal interface at `/webterminal` with interactive `clickhouse-client` over WebSocket.
- `Kafka2` engine now supports direct `SELECT` queries and `kafka_commit_on_select` setting.
- `WASM` UDFs support more numeric type coercions.
- Support `LIKE` function in `DELETE FROM system.webassembly_modules`.
- Support geo types for `Iceberg`.
- Add prepared statements functionality to ArrowFlight SQL server.
- Improved KQL parser robustness.

### Performance Improvements

- Reuse Parquet footer metadata cache for local files.
- Push `ORDER BY ... LIMIT n` through `LEFT`/`RIGHT` joins when applicable.
- Enable `use_top_k_dynamic_filtering` and `use_skip_indexes_for_top_k` by default.
- Use jemalloc oversize-arena feature to reduce page faults.
- Limit simultaneously active streams in `UNION ALL` to reduce peak memory.
- Optimize userspace page cache; cold reads through object storage page cache now coalesce consecutive misses into single HTTP request.
- Speed up index analysis with type casting and function application.
- Speed up queries on `Merge` tables with many underlying tables.
- Remove vtable from settings, reducing `Settings` copy size by ~28x.
- Added `max_threads_min_free_memory_per_thread` and `max_insert_threads_min_free_memory_per_thread` settings for automatic parallelism reduction.
- Reduce memory usage on low-memory systems (<4 GiB).
- Added `OptimizeTrivialGroupByLimitPass` for trivial GROUP BY queries.
- Granule-level implicit min-max index for `_part_offset` and `_block_number` virtual columns.
- Precompute per-row hashes for `prealloc_serialized` aggregation methods.
- Query condition cache for `Iceberg` tables.
- Optimize `cramersV` and related functions with window functions.
- Rewrite `tupleElement(dictGet(...), N)` optimization.
- Enable buffering for sorting steps in distributed queries.
- Performance improvements for `partial_merge` JOIN.
- Reduced memory over-allocation in `partial_merge` JOIN.
- Optimize memory tracker sampling.
- Reduce memory allocation during S3 multipart uploads.
- Add software prefetch in hash join probe phase.
- Optimize `MemoryTracker` layout improving performance by ~25%.
- Rewrite `coalesce(a, b, ...) <op> const` predicates for better index pruning.
- Significantly improved Iceberg catalog query performance with large JSON metadata.
- Avoid index uncompressed cache overhead when disabled.
- Fix excessive granule reading in `ORDER BY ... LIMIT` with non-selective filters.
- Fixed missing hash table sizes cache reuse after JOIN reordering.
- Fsync of part files now performed in parallel, speeding up insert finalization by ~22%.
- Vectorize string search helpers on AArch64 using NEON.
- Restore I/O savings for aliased complex columns.
- Reduced lock contention on asynchronous remote-FS read path.
- Slightly better query plan for projections in-order scan.
- Extend projection optimization to handle cases where projection's sorting key removes sorting step.
- Allow `json.path[]` syntax sugar on typed JSON paths.
- Optimize `json.path[N].nested.path` expansion.
- Optimize `SHOW TABLES` access checks.
- Remove redundant per-table access check in `system.tables` fast path.
- Enable JIT compilation on macOS.
- Skip row policy deferral after `FINAL` when deterministic and depends only on sorting-key columns.
- Support `hasAny` and `hasAll` as text index filter predicates.
- Route long-lived MergeTree heap state to dedicated jemalloc arena.
- Improve ZooKeeper client timeout handling under heavy load with progress-based timeout.
- Move per-event trace flags into separate array with lazy allocation.
- Add `compareTrackAt` for `ColumnDecimal`.
- Added SIMD implementation of `MD5` (AVX2/AVX512) for parallel hashing.

### Improvements

- Improved `--help` output for all ClickHouse applications with better formatting and subcommand listing.
- Changed default `input_format_column_name_matching_mode` from `match_case` to `auto` for case-insensitive fallback.
- Added `STDDEV` alias of `stddevSamp`.
- Added `array_to_string` alias of `arrayStringConcat`.
- Added `unnest` alias of `arrayJoin`.
- Progress bar in `clickhouse-client` now shows temporary disk data usage.
- Added `system.predicate_statistics_log` for predicate filter selectivity sampling.
- Allow skipping local shard with missing table when `skip_unavailable_shards` enabled.
- Added startup warnings for Linux mdraid array issues.
- WASM UDFs now appear in `system.functions` with correct metadata.
- `generate_series` table function supports negative step values.
- Fixed context expiration in functions used in deferred execution paths.
- Fixed spurious cache limit violations.
- Auto-detect region for S3 Express endpoints.
- Support field IDs for `Iceberg` data files.
- Left panel in web UI scrolls independently and stays visible.
- Allow opening Play UI in new tab via Ctrl/Cmd/Shift+click.
- New metrics for projection primary key and index granularity memory usage.
- Apply network bandwidth limits to remote filesystem operations.
- Improved backup resilience for Refreshable Materialized Views.
- Consistent database listing in `system.databases` regardless of catalog settings.
- Aligned `concurrent_threads_soft_limit_ratio_to_cores` default in code with config.
- Improved sanitizer robustness in parser and string functions.
- Improved monotonicity inference for `divide(0, x)` and `intDiv(0, x)`.
- Replaced HTTPConnectionPool async metrics with bucket-based histogram.
- Socket buffer settings now reported as changeable without restart.
- Added symbol resolution support in `flameGraph`.
- CLI client supports `<rainbow_parentheses>false</rainbow_parentheses>` config.
- Fixed `MultiVolnitsky` UTF-8 case-insensitive search.
- Implemented `sched_getcpu` via `rseq` TLS in glibc-compatibility musl.
- Replicated refreshable materialized views no longer refresh twice on ZK reconnection.
- Allow Distributed tables without explicit column list to validate sharding keys.
- Fixed REST catalog with Azure `abfss` paths.
- Honor `role_arn` and `role_session_name` in Keeper S3 snapshot client.
- Fixed memory leak in `MATERIALIZED` CTEs.
- Fixed exception in `lowerUTF8`/`upperUTF8` with large non-ASCII datasets.
- Hardened MergeTree column-statistics loading against transient I/O failures.
- `INSERT INTO ... SELECT FROM input(...)` no longer requires `CREATE TEMPORARY TABLE` grant.
- Added `concurrent_part_removal_threshold_for_remote_disk` setting (default 16).
- New setting `defer_partition_pruning_after_final` for opt-out of 26.3 behavior.
- Fixed `intExp2` returning wrong results for out-of-range inputs.
- Fixed shell argument quoting in `clickhouse install` and `clickhouse git-import`.
- `Alias` table engine is now non-experimental.
- Added `--queries-format` option to `clickhouse-benchmark`.
- Tuple element-wise operators are now variadic.
- Added `/processors-profile` web UI for pipeline visualization.
- Setting `query_plan_use_logical_join_step` is now obsolete.
- Pre-filled `user` input removed from web UI pages.
- Decoupled `ORC` reader offset-based read from `use_prefetch`.
- Downloaded CSV files now use `.csv` extension.
- More input validation for BigLake catalog.

### Bug Fixes

- Fixed inconsistent expression formatting with parenthesis tracking.
- Fixed accurate comparison between Decimal and Float types.
- Allow positional arguments in distributed queries.
- Fixed assertion failure when dropping/recreating materialized views and view dependencies.
- Fixed server abort when creating table with duplicate UUID.
- Fixed skip index usage on data read with patch parts from lightweight updates.
- Fixed "Not-ready Set" exception with IN subqueries moved to PREWHERE.
- Fixed "Cannot find column" error combining ADD and RENAME COLUMN.
- Fixed views with mixed set operations returning wrong results after detach/attach.
- Fixed PLACEHOLDER action exception in stress testing.
- Fixed logical error with Iceberg Date partitions.
- Fixed skip index incompatibility after column type modification.
- Fixed credential leak in query logs.
- Fixed DROP TABLE on Kafka tables potentially hanging indefinitely.
- Fixed runtime lookup for source table engines and database source grants.
- Fixed server crash with ALIAS column containing correlated subqueries.
- Fixed server crash with correlated subqueries and CONSTRAINT ASSUME.
- Fixed exception with JSON dynamic fields as GROUP BY key.
- Fixed workload IO scheduling bypass for S3/object-storage writes.
- Fixed Protobuf/ProtobufList schema cache collision.
- Fixed heap-buffer-overflow in Rust CXX bridge.
- Fixed FunctionVariantAdaptor exception handling.
- Fixed TRUNCATE ALL TABLES with views.
- Fixed `optimize_rewrite_array_exists_to_has` type compatibility and re-enabled.
- Fixed S3 settings priority for storage configuration.
- Fixed sparse serialization losing negative zero sign.
- Fixed WKT geometry parser MULTILINESTRING corruption and malformed string handling.
- Fixed GROUPING SETS on distributed single-shard tables.
- Fixed LOGICAL_ERROR column identifier already registered.
- Fixed LOGICAL_ERROR with `additional_result_filter` on UNION/EXCEPT.
- Fixed server crash with * APPLY on aggregate functions with grouping.
- Fixed infinite loop with CSV/TSV skip first lines exceeding file lines.
</Release>

<Release version="26.4" date="April 30, 2026" published="2026-04-30T00:00:00.000Z" url="https://clickhouse.com/docs/whats-new/changelog#26-4">
## ClickHouse release 26.4

## Backward Incompatible Changes

- The `IN` operator now uses exact value semantics for `Bool` type: only `0` and `1` values in the set match `Bool` values. Previously, numeric values greater than `255` in the `IN` set were incorrectly clamped to true.
- The H3 library has been updated to v4, which improves the precision of length, area, and other metric calculations. This change is backward incompatible because the new results differ from previous ones.
- Disallows using `SELECT` as a bareword identifier in a `WITH` expression list element.
- The merge table will now handle virtuals differently: if the underlying table contains `_table` or `_database`, these columns will be read from storage; otherwise, they will be filled after the read step.
- The `IN` operator now rejects lossy `Decimal` conversions inside composite types (`Tuple`, `Array`, `Map`).
- Reduced default `http_max_fields` from 1,000,000 to 1,000 and `http_max_field_name_size` from 128 KB to 4 KB to limit pre-authentication memory usage by HTTP connections.

## New Features

- Automatic spilling to hash and parallel hash joins by converting them to grace hash join when memory limit is reached.
- Arrow Flight SQL support.
- Incremental read support for `Paimon` table engines with Keeper-backed snapshot progress tracking.
- The `stem` function is now non-experimental.
- New behavior of `max_insert_block_size_rows` and related settings under compatibility setting `use_strict_insert_block_limits`.
- Function `arrayAutocorrelation(arr [, max_lag])` for computing normalized autocorrelation of numeric arrays.
- SQL function `obfuscateQuery` for query obfuscation.
- Support for Map and JSON/Object types as dictionary attributes.
- New MergeTree settings `replicated_fetches_min_part_level` and `replicated_fetches_min_part_level_timeout_seconds` to reduce replication overhead.
- MergeTree skip index support for JSON columns using JSONAllPaths.
- The `printf` function now supports non-constant format strings.
- New projection index `commit_order` that reorganizes data in insertion order.
- Function `highlight` that wraps search terms in HTML tags.
- Quotas by normalized query hash to protect public ClickHouse services.
- Support for `NATURAL JOIN` syntax.
- Support `SET TIME ZONE 'tz'` as alias for `SET session_timezone`.
- Parameterized queries in the Web UI with input fields for parameter values.
- SQL standard `VALUES` clause as a table expression.
- PostgreSQL-compatible units for `EXTRACT` operator: `EPOCH`, `DOW`, `DOY`, `ISODOW`, `ISOYEAR`, `WEEK`, `CENTURY`, `DECADE`, `MILLENNIUM`.
- SQL-standard compound interval literals with `TO` range qualifiers.
- Asynchronous metrics for kernel TCP receive and transmit buffer memory.
- Jemalloc profiling web UI for ClickHouse Keeper.
- Command `SYSTEM FLUSH OBJECT STORAGE QUEUE db.table PATH 'x'`.
- Function `JSONAllValues` returning all values from JSON columns as Array(String).
- Setting `input_format_column_name_matching_mode` for different case sensitivities in input formats.
- `watch` command in `clickhouse-keeper-client`.
- `getChildrenRecursive` (ListRecursive) request to ClickHouse Keeper.
- Function `arrayTranspose` for transposing two-dimensional arrays.
- Auto statistics defaults: `auto_statistics_types mergetree` defaults to `'minmax, uniq'`; `materialize_statistics_on_insert` defaults to false.
- Setting `prefer_dependency_replica` for materialized view dependency chains.
- Function `hasPhrase` (alias `matchPhrase`) for phrase search.
- Histogram metrics `s3_read_request_duration_microseconds` and `s3_read_request_bytes`.
- `Date` and `Date32` values can now be added to `Time` and `Time64` values using the `+` operator.
- Text index is now GA and stays enabled regardless of the `compatibility` setting.

## Experimental Features

- `ALTER TABLE ... EXECUTE remove_orphan_files` for Iceberg tables.
- Setting `query_plan_optimize_join_order_randomize` for randomizing join reordering statistics.
- AI function support: `aiGenerate`, `aiClassify`, `aiExtract`, `aiTranslate` for calling OpenAI and Anthropic endpoints.
- System table `system.histogram_metric_log` for periodically snapshotting histogram metrics.

## Performance Improvements

- ClickHouse can now prune entire data parts based on min/max statistics.
- Reduced lock contention during readonly operations on ReplicatedMergeTree tables.
- Respect `optimize_read_in_order` when reading projections.
- Improvements in Hash Join and Concurrent Hash Join.
- Optimize `DISTINCT` by disabling `LowCardinality` optimization when input is almost distinct.
- Performance optimization for `LIKE` queries using text indices.
- Vectorized math functions accelerated on AArch64 and FreeBSD/Darwin.
- Primary key pruning for regexp alternations over literals with common prefixes.
- Generalized `ORDER BY ... LIMIT` top-k dynamic filtering for `Nullable`, `String`, `COLLATE` types.
- Faster hash join on `Int32` and `Int64` keys with small range.
- Faster discontinuous queries for `LowCardinality` columns.
- Faster `var*Stable` and `stddev*Stable` functions for `Float64` columns.
- Optimized base58 encode/decode operations.
- Binary size reduction and instruction cache utilization via linker optimizations.
- Fixed negative scaling for short queries with aggregation.
- Improved performance with parallel replicas.
- Prefetching support for remote file reads.
- Avoided unnecessary String `.size` subcolumn computation.
- Less jittery progress bar in clickhouse-client.
- MemoryWorker support in clickhouse-local for userspace page cache.
- `LIMIT` clause push-down optimization for `UNION ALL`.
- JIT compilation for `String` and `FixedString` column comparisons in `ORDER BY`.
- Virtual row boundary information emission for improved merge reprioritization.
- Faster Float-to-String conversion using extended `itoa` and `zmij` library.
- Faster `Int128`/`UInt128` to string conversion.
- Avoided redundant threads in `uniqExact` parallel merge; batch parallel merge support.
- Better parallelization of simple views with parallel replicas.
- Parallel replicas support over simple views including `UNION ALL` views.
- Optimized reading in order with `IN` filters in `full_sorting_merge`.
- Cached sampling settings to optimize allocations/deallocations.
- Fixed INSERT performance regression with `deduplicate_insert = 'enable'`.
- Reduced profiled lock overhead.
- AVX2 and ARM NEON support for `arrayDotProduct`.
- Improved `INSERT VALUES` performance for `Map`, `Array`, `Tuple` types.
- Fixed excessive RabbitMQ table engine CPU usage.
- JOIN order optimizer now infers transitive equi-join predicates.
- `TRUNCATE DATABASE TABLES LIKE` optimization with parallel merge cancellation.
- Monotonicity support for multiply in primary key pruning.
- Cache dictionaries no longer take exclusive lock in `hasKeys`.
- Inlined VIEW subqueries for more optimizations.
- Optimized cache loading on server startup.
- Lazy column materialization for ReplacingMergeTree with FINAL.
- Re-enabled `optimize_rewrite_array_exists_to_has` optimization with compatibility fixes.

## Improvements

- Improved EXPLAIN PLAN output with better formatting and join information.
- Added MergeTree setting `share_nested_offsets` for independent nested column handling.
- Multiple authentication methods in users.xml/yaml configuration.
- Auto reload Raft inter-node connections using TLS.
- Extended `cast_keep_nullable` to work with Dynamic/JSON types.
- Reduced memory footprint from internal ISerialization objects via object pool.
- Support for `password` and `identity` fields in keeper-client XML config.
- Improved Iceberg writes for unity catalog.
- Setting `finalize_projection_parts_synchronously` for synchronous projection finalization.
- Added `projections_duration_ms` column to `system.part_log`.
- Improved query cancellation via KILL QUERY and clickhouse-client Ctrl+C.
- Replaced hardcoded `source_table_engines` with runtime lookup.
- Setting for type mismatch behavior in Variant and Dynamic.
- Improved Iceberg and Spark compatibility with consistent path handling.
- Fixed race condition in `IPartitionStrategy::cached_result`.
- ClickHouse Interval datatypes can now be written in Arrow Format.
- Native support for UUID data types in Arrow and Parquet formats.
- Support for 7z archives on object storage.
- ProfileEvents for object storage introspection.
- Fixed `merge` table function with non-present columns.
- Included commit time in mutation execution time metric.
- Write-ahead log for blob objects pending removal.
- Disabled AI SQL generation in embedded client for security.
- Changed Iceberg insert interface with deprecated settings.
- Tabs rendered as 4 spaces in clickhouse-client.
- Avoid scanning remote data lake catalog for table hints when disabled.
- Applied `distributed_index_analysis_min_indexes_bytes_to_activate` after partition pruning.
- Fixed Parquet bloom filter push-down assertion with empty IN/NOT IN clauses.
- MinMax column statistics now store values as Field with type name serialization.
- Updated cppkafka with Consumer close deadlock fix.
- Iceberg file object information now includes row count and size from manifest.
- Configuration parameter `use_separate_cache_arena` for cache memory arena separation.
- Native support for Arrow `StringView` and `BinaryView` data types.
- Hot-reloaded Keeper settings: `max_requests_batch_size`, `max_requests_batch_bytes_size`, etc.
- Incremented profile events `MemoryAllocatedWithoutCheck` in release build.
- Cgroupv2 memory tracking excludes `slab_reclaimable`.
- `use_partition_pruning = 0` disables MinMax index pruning and count optimization.
- `pretty=1` in EXPLAIN prints expressions in human readable format.
- `accurateCastOrNull` and `accurateCastOrDefault` support `Tuple` target types.
- Fixed chart duplication in Play UI theme switching.
- Updated chdig tool to newer versions with perfetto UI and other improvements.
- Trailing comma support in `WITH` clause before SELECT.
- Setting `compress_per_column_in_compact_parts` for compressed block organization.
- Improved Play UI with better table info balloon and engine-specific icons.
- Support for `Nullable(Tuple)` in Arrow, ArrowStream, ORC, Parquet formats.
- TOTALS row displayed as table footer in web UI.
- Multi-query mode in web UI with parallel SELECT execution.
- Fixed column resize in web UI result table.
- Limit on jemalloc profile flushes per time interval.
- Keeper settings hot-reload: `nuraft_streaming_mode`, `nuraft_max_log_gap_in_stream`.
- `CGroupMemoryUsedWithoutPageCache` async metric.
- Parser-level syntactic sugar for SQL standard `OVERLAY` function.
- Column alias `INDEX_LENGTH` in `information_schema.tables`.
- `information_schema.tables` now ignores inactive table parts.
- `ngrams` function now rejects invalid ngram lengths.
- FIPS-specific test filtering and improvements.
- Expandable cells in Web UI with three-line height limit.
- Option to force virtual/path style for S3 endpoints.
- `restore_replace_external_engines_to_null` now skips external engine databases.
- Text index analysis support for `hasPhrase` function.
- STATISTICS treated as read-only in ColumnDependency.
- `system.asynchronous_metric_log` creation in keeper-as-server mode.
- Configuration option `default_system_log_flush_policy.skip_alias_columns`.
- Auto statistics disabled for system tables.
- `array` tokenizer support for LIKE optimization.
- MemoryAllocatedWithoutCheck sent in release builds.
- Per-thread untracked_memory exposed in `system.stack_trace`.

## Bug Fixes

- Fixed `Block structure mismatch in stream` error from Lazy materialization.
- Fixed logical error with data masking policy query on CLUSTER.
- Fixed bug when using Unity catalog on GCS.
- DataLakeCatalog respects server's `http_forbid_headers` configuration.
- Fixed N+1 `HeadObject` calls for S3 brace-expansion globs.
- Validate setting changes in create queries.
- Fixed `ALTER TABLE UPDATE/DELETE` with MATERIALIZED columns depending on EPHEMERAL columns.
- Credentials in JDBC, ODBC, NATS connection strings are now masked.
- Fixed `parseDateTimeBestEffort` incorrectly parsing words with month prefixes.
- Fixed ignoring TABLE_UUID_MISMATCH for non-analyzer.
- Fixed explicit settings ignored when matched server default with compatibility setting.
- Fixed numbers with leading zeros in hive partitioning causing errors.
- Fixed heap-use-after-free when table dropped concurrently with read query.
- Fixed Keeper read request stuck when unrelated session closed at wrong moment.
- Validate column structure before applying patches.
- Fixed vertical merge assertion with Dynamic columns and SYSTEM STOP/START MERGES.
- Fixed incorrect partition pruning for `toWeek()` function.
- Fixed exception in functions with ColumnReplicated from JOIN.
- Fixed `CLEAR COLUMN` not rebuilding projections and materialized columns.
- Fixed exception in `ConditionSelectivityEstimator` with scalar query parameters and statistics.
- Fixed parts with unknown projections marked as lost forever.
</Release>

<Release version="26.3" date="March 26, 2026" published="2026-03-26T00:00:00.000Z" url="https://clickhouse.com/docs/whats-new/changelog#26-3">
## ClickHouse release 26.3 LTS

#### Backward Incompatible Changes

- **Data type serialization:** Propagate data types serialization versions to nested data types. String serialization version `with_size_stream` now applies to String types inside Array/Map/Variant/JSON/etc. Controlled by `propagate_types_serialization_versions_to_nested_types` (enabled by default). **Upgrade is safe, but downgrade is not** — data written by 26.3 in columns with nested types will be unreadable in older versions.
- **Skip index type:** Removed the `hypothesis` skip index type. Creating tables with `INDEX ... TYPE hypothesis` now produces an error.
- **Function removal:** Removed experimental `detectProgrammingLanguage` function.
- **NOT operator precedence:** Fixed to match SQL standard. `NOT` now binds looser than `IS NULL`, `BETWEEN`, `LIKE`, and arithmetic operators. Queries relying on previous non-standard behavior may change results.
- **Skip index filenames:** Fixed skip index files not respecting `replace_long_file_name_to_hash` setting. Long-named indices are now hashed like column files. Backward compatible for upgrades, but downgrading may cause long-named indices to be ignored.
- **Async insert default:** Async insert is now enabled by default. Batching all small inserts by default. Can be disabled at config, session, query, or table level. Set `compatibility=<version less than 26.2>` for previous behavior.
- **MySQL datatype mapping:** Changed default `mysql_datatypes_support_level` to `decimal,datetime64,date2Date32`, enabling proper mapping of MySQL DATE/DECIMAL/DATETIME to Date32/Decimal/DateTime64. Previously MySQL DATE mapped to Date (cannot represent dates before 1970-01-01).
- **UDF stderr handling:** Changed default `stderr_reaction` from `throw` to `log_last` for executable UDFs. UDFs writing warnings to stderr no longer fail when exit code is 0.
- **Metadata changes:** Fixed normal projections metadata so multi-column sorting keys are properly recognized.
- **Index function API:** `mergeTreeAnalyzeIndexes{,UUID}` now accepts array of part names instead of regexp for improved performance (experimental).

#### New Features

- **Bucketed Map serialization:** Added bucketed serialization for Map columns in MergeTree with `map_serialization_version = 'with_buckets'`. Keys split into hash-based buckets, providing 2-49x speedup for single-key lookups. New settings: `map_serialization_version`, `max_buckets_in_map`, `map_buckets_strategy`, `map_buckets_coefficient`, `map_buckets_min_avg_size`.
- **Materialized CTE:** Support evaluating CTEs only once and storing results in temporary tables.
- **SQL-standard functions:** Allow certain functions like `NOW` without parentheses for compatibility.
- **naturalSortKey function:** Added `naturalSortKey(s)` function.
- **JSON input for JSONExtract:** Support native JSON/Object input for JSONExtract functions.
- **Nullable query parameters:** If a query parameter has Nullable type and is not specified, assume value is NULL.
- **ZooKeeper auxiliary support:** Support auxiliary ZooKeeper for Replicated database.
- **JSON has function:** Support `has` function for JSON type to check path existence.
- **mergeTreeTextIndex function:** New table function to read data directly from a text index for introspection.
- **table_readonly setting:** Add MergeTree setting to mark tables as read-only.
- **Partition pruning settings:** New `use_partition_pruning` setting and `use_partition_key` alias to disable partition pruning.
- **Iceberg EXECUTE expire_snapshots:** Support `ALTER TABLE ... EXECUTE expire_snapshots('<timestamp>')` for Iceberg tables.
- **HTTP handlers per port:** Allow each `type=http` entry in `<protocols>` to specify custom `<handlers>` key for different routing rules per port.
- **EXPLAIN improvements:** Add `pretty=1` option for tree-style indented output and `compact=1` to collapse Expression steps.
- **Access control:** New `restore_access_entities_with_current_grants` setting for restored users/roles in backups.
- **Unicode functions:** Added `caseFoldUTF8`, `removeDiacriticsUTF8`, and `normalizeUTF8NFKCCasefold` functions.
- **asciiCJK tokenizer:** New tokenizer for full-text indexes and `tokens` function using Unicode word boundary rules.
- **Unavailable shards limits:** New `max_skip_unavailable_shards_num` and `max_skip_unavailable_shards_ratio` settings to limit silent shard skipping.
- **SOME keyword:** Added `SOME` keyword for subquery expressions (identical to `ANY`).
- **FixedString output formatting:** New `output_format_trim_fixed_string` setting to strip trailing null bytes.
- **Parenthesized table joins:** Support parenthesized table join expressions in FROM clause.
- **toDaysInMonth function:** Returns number of days in the month of specified date.

#### Experimental Features

- **WebAssembly UDFs:** Experimental support for WebAssembly-based user-defined functions with Wasmtime backend.
- **External SQL dialects:** Support for external SQL dialects using `polyglot` library.
- **ALP codec:** Floating-point compression codec (without ALP_rd fallback).
- **Lazy JSON type hints:** Experimental lazy type hints for JSON columns. `ALTER TABLE ... MODIFY COLUMN json JSON(path TypeName)` completes as metadata-only operation without rewriting data.
- **YTsaurus parallel reads:** Enable parallel reads from YTsaurus table engine.

#### Performance Improvements

- **Object storage reads:** Improved data lake reading performance by resizing pipeline to number of processing threads, providing ~40x improvements on multi-core machines.
- **Parallel replicas logic:** Clarified relationship between `enable_parallel_replicas` and `automatic_parallel_replicas_mode`. Queries use parallel replicas if `enable_parallel_replicas > 0`. `automatic_parallel_replicas_mode=1` uses statistics-based decision; mode=0 uses parallel replicas for all supported queries.
- **Enhanced partition pruning:** Allow partition pruning when predicate contains comparison operators and partition key is wrapped in deterministic function chain (e.g., `cityHash64(x) % 5 > 2`, `toYYYYMM(x) < 2026`).
</Release>

<Release url="https://clickhouse.com/docs/whats-new/changelog#bug-fixes-and-improvements">
## Bug Fixes and Improvements

- Fixed `sumCount` aggregate function unable to read older serialized states after introduction of `Nullable(Tuple)`
- Fixed exception in tuple comparison involving `Nothing` type with `GROUPING SETS` and `ORDER BY`
- Fixed non-deterministic `uncompressed_hash` computation for Compact MergeTree parts with multiple compression codecs
- Fixed logical error during INSERT SELECT with JSON and buckets in shared data
- Fixed `MEMORY_LIMIT_EXCEEDED` being reported as `CORRUPTED_DATA` during SummingMergeTree and CoalescingMergeTree merges
- Fixed "Context has expired" exception for correlated subqueries with table functions like `url()`
- Fixed exceptions and incorrect behavior in `optimize_syntax_fuse_functions` with aggregate projections and Date types
- Fixed `replaceRegexpOne` to `extract` query rewrite producing wrong results; fixed exception with `GROUP BY ... WITH CUBE` and `group_by_use_nulls=1`
- Fixed `DROP DATABASE` hanging indefinitely with `database_atomic_wait_for_drop_and_detach_synchronously`
- Fixed `KILL QUERY` unable to terminate queries stuck in `WITH FILL`, dictionary loading, or `ALTER DELETE`
- Fixed security issue: `loop` table function bypassed row policies and column-level grants
- Fixed incorrect partition pruning with pre-epoch DateTime64 and `toDate()` function
- Fixed `hasPartitionId` returning incorrect results with multiple partitions
- Fixed possible crashes reading empty granules in advanced shared data in JSON
- Fixed `Cannot schedule a file` error on `INSERT` into `Distributed` due to race between `DROP` and `INSERT`
- Fixed server crash in `mapContainsKey/mapContainsKeyLike` with `tokenbf_v1` skip index
- Fixed `LOGICAL_ERROR` with `LowCardinality` inside compound types in various functions
- Fixed `LOGICAL_ERROR` in `MergingAggregatedTransform` with `ARRAY JOIN` and `merge()` over Distributed tables
- Fixed server crash caused by uncaught exception in HTTP connection pool destructor
- Fixed incorrect result with `grace_hash` algorithm and non-equi joins
- Fixed performance inefficiency in DeltaLake metadata scanning
- Fixed data race in ZooKeeper client between sendThread and receiveThread
- Fixed bug preventing CTE with distributed insert selects
- Fixed exception from `CachedOnDiskReadBufferFromFile::readBigAt`
- Fixed `LOGICAL_ERROR` in `Alias` engine with materialized columns
- Fixed Keeper data loss after restart with Azure Blob Storage
- Fixed JIT miscompilation of `sign` function for integer types wider than `Int8`
- Fixed `DUPLICATE_COLUMN` exception and silent NULLs reading Delta Lake tables with name mode column mapping
- Fixed mutation after lightweight update and secondary indices
- Fixed incorrect result of FINAL queries mixing primary and non-primary key skip indexes
- Enforced READ ON FILE checks for scalar `file()` and `DESCRIBE TABLE file()`
- Fixed crash with glob pattern in `file()` when directory contains dangling symlinks
- Fixed segfault in query plan optimization converting outer join to inner join with `arrayJoin`
- Fixed `ProtobufList` format not working with Kafka engine
- Fixed logical error with `analyzer_compatibility_join_using_top_level_identifier` and ARRAY JOIN
- Set `Watch` component for watch responses in `aggregated_zookeeper_log`
- Fixed partition pruning with FINAL deduplication when partition key not covered by sorting key
- Fixed logical error in `kql_array_sort_asc`/`kql_array_sort_desc` with constant array arguments
- Fixed out-of-bounds access in `ColumnConst::getExtremes` with `extremes = 1`
- Fixed potential deadlock with concurrent `MOVE PARTITION` operations
- HTTP server now returns error message in response body for 400 Bad Request
- Fixed wrong results with distributed index analysis and query condition cache
- Fixed `LOGICAL_ERROR` in `MergeTreeSetIndex` with `toDate` conversion on key columns
- Fixed `LOGICAL_ERROR` with RIGHT JOIN wrapped in CROSS JOIN in legacy join code path
- Fixed segfaults and OOM from corrupted `DDSketch` deserialization in `quantilesDD`
- Fixed `LOGICAL_ERROR` with correlated columns in lambda functions like `arrayMap`
- Fixed logical error in `caseWithExpression` with `materialize(NULL)` or `Nullable(Nothing)` arguments
- Fixed bad cast exception when filtering `_table` virtual column in `merge` table function
- Fixed sporadic deduplication failure with inconsistent cleanup ordering in ZooKeeper
- Fixed exception with `ORDER BY ... WITH FILL` used together with `LIMIT BY`
- Fixed silent data corruption inserting Parquet/Arrow `Date` into `Enum` column
- Fixed exception reading Arrow file with `Array` column into table with `Nested` column
- Fixed `MATERIALIZE INDEX` and `MATERIALIZE PROJECTION` mutations getting stuck
- Fixed exception reading from `Nullable(Tuple(...))` with colliding element names
- Fixed exception when joining `Merge` table wrapping `Distributed` table
</Release>

<Release url="https://clickhouse.com/docs/whats-new/changelog#clickhouse-bug-fixes-release">
## ClickHouse Bug Fixes Release

Multiple bug fixes including:

**Parquet and Data Format Fixes:**
- Fixed incorrect Parquet Bool to FixedString conversion in native V3 reader
- Fixed HTTP Basic Auth to accept base64 credentials without padding

**Query Execution Fixes:**
- Fixed incorrect partition pruning with Nullable partition key columns
- Fixed rare exception in pipeline executor during query cancellation
- Fixed 'Column identifier is already registered' exception with count_distinct_optimization and QUALIFY
- Fixed exception with IN/NOT IN using LowCardinality column arguments
- Fixed 'Pipeline stuck' exception in full_sorting_merge joins
- Fixed LOGICAL_ERROR with TTL merging and aggregate projections
- Fixed logical error with CROSS JOIN and INNER JOIN USING
- Fixed null pointer dereference in dictGetOrDefault with Nullable keys
- Fixed exception in DISTINCT queries with aggregate projections and LowCardinality
- Fixed LOGICAL_ERROR with arrayJoin in filter expression with OUTER JOIN
- Fixed logical error with parallel replicas and optimize_aggregation_in_order
- Fixed pipeline deadlock with sort_overflow_mode and window functions
- Fixed column rollback in Buffer engine during exception handling

**Type System and Comparison Fixes:**
- Fixed bad cast exception in null-safe comparison with Dynamic/Variant and NULL
- Fixed IS DISTINCT FROM with Dynamic/Variant vs NULL returning incorrect results

**Index and Optimization Fixes:**
- Fixed tryGetColumnDescription to filter subcolumns by parent column kind
- Fixed text index usage with other skip indexes
- Fixed distributed index analysis with expressions in primary key
- Fixed S3 requests being incorrectly retried on non-retryable errors
- Fixed set skip index usefulness detection with OR predicates

**ClickHouse Keeper Fixes:**
- Fixed Keeper disconnecting Java ZooKeeper clients after addWatch request
- Fixed zk_followers and zk_synced_followers metrics not decreasing
- Fixed Keeper's secure raft port ignoring cipherList and dhParamsFile configuration
- Fixed misleading Keeper log messages about operation timing
- Fixed Keeper TCP connections preventing graceful server shutdown

**Join and Filter Fixes:**
- Fixed LOGICAL_ERROR with PREWHERE and IN subquery on MergeTree tables
- Fixed exception 'Sorting column wasn't found in ActionsDAG' with query_plan_convert_join_to_in
- Fixed BAD_GET exception with non-boolean expression in WHERE and SELECT with JOIN
- Fixed LOGICAL_ERROR with read_in_order_through_join and parallel replicas

**Data Type and Function Fixes:**
- Fixed MongoDB dictionary source failing with named collections
- Fixed LOGICAL_ERROR when Identifier is empty after parameter substitution
- Fixed exception with input table function as argument of remote
- Fixed outdated data parts resurrection from incorrect cleanup
- Fixed exception in LogicalExpressionOptimizerPass with Variant return type
- Fixed parseDateTimeBestEffort incorrectly parsing month/weekday prefixes
- Fixed UNKNOWN_IDENTIFIER with merge() table function over tables with different JSON parameters
- Fixed optimize_skip_unused_shards with Distributed storage in Views
- Fixed tuple subcolumn access by name for external tables
- Fixed reverseUTF8 exception on invalid UTF-8 input
- Fixed LOGICAL_ERROR with FINAL, PREWHERE, constant WHERE and column-independent aggregates

**Configuration and System Fixes:**
- Fixed system.trace_log entries for dictionary auto-reloads having non-empty query IDs
- Fixed crash from null pointer dereference in system tables iteration
- Fixed SYSTEM START REPLICATED VIEW not waking up refresh task
- Fixed 'Inconsistent table names' exception with view() containing JOINs
- Fixed adjusting RLIMIT_SIGPENDING via pending_signals

**Security Fix:**
- Fixed RBAC bypass allowing users to DESCRIBE any table via remote/cluster functions without SHOW_COLUMNS privilege

**Performance Fixes:**
- Fixed unexpected result with read_in_order_use_virtual_row and monotonic functions
- Fixed JIT expression compilation bugs preventing optimization passes

**TTL and Transactions:**
- Fixed LOGICAL_ERROR in renameAndCommitEmptyParts with concurrent TRUNCATE TABLE and OPTIMIZE TABLE
- Fixed decimal overflow when partition pruning with DateTime64
</Release>

<Release url="https://clickhouse.com/docs/whats-new/changelog#clickhouse-release">
## ClickHouse Release

Add information about deferred filters as a separate item to EXPLAIN query output. Enable `type_json_allow_duplicated_key_with_literal_and_nested_object` by default to avoid errors about duplicated keys during JSON parsing. Keeper improvement: `find_super_nodes` command is more reliable when finding multiple super nodes by forbidding traversal of their children. Initial completion support for `clickhouse-keeper-client`. Flush async logging buffers in case of crash. Enable the impersonate feature by default (see EXECUTE AS target_user). Improve canceling queries with SQLite, MongoDB and MySQL table engines by KILL QUERY and cancel query (Ctrl+C). Add server setting `jemalloc_profiler_sampling_rate` to control jemalloc profiling. Support weights in concurrent bounded queue implementation. Add sslmode to allowed keys for PostgreSQL dictionary sources to support SSL connections. Show clear "no such file" error when passing non-existent file paths. Text indexes can now be built on Nullable and Array(Nullable) columns. Avoid dropping named collections that are dependencies of dictionary sources. Enable `grace_hash` join algorithm for queries with totals. Cancel background merges early in DROP DATABASE. Remove NetlinkMetricsProvider and use procfs exclusively for per-thread taskstats metric collection. Refactor Iceberg manifest file handling to fix caching issues. Optimize PREWHERE decisions for expressions like `toDate(time)`. Add `MaxAllocatedEphemeralLockSequentialNumber` metric for ZooKeeper. Add new profile event `KeeperRequestTotalWithSubrequests` for better Keeper workload visibility. `SYSTEM RELOAD DICTIONARIES` now reloads dictionaries in topological order. Restart statistics cache after changing MergeTree settings. Only "alive" replicas participate in distributed index analysis. Add setting `access_control_improvements.disallow_config_defined_profiles_for_sql_defined_users`. Cap automatic parallel replicas heuristic to actual node count. Implement hedged requests and asynchronous reading for distributed index analysis. Deserialization of binary `AggregateFunction` states now requires consuming full input. Make TRUNCATE DATABASE respond to query cancellation. Improve `keeper-bench` with request pipelining and better stats. Support SAMPLE clause in distributed index analysis. Show chart title in dashboard even with empty results. Cap column lists in analyzer error messages to 10 entries. Return column stats from sub-queries with joins. Mark ZooKeeper session as expired immediately on finalization. Use more math functions from LLVM-libc. Reduce memory usage in `system.jemalloc_profile_text`. Add `is_subrequest` column to `system.aggregated_zookeeper_log`. Allow `ALTER TABLE MODIFY COLUMN x TTL` without specifying column type. Skip stale Keeper requests for disconnected sessions. Support text index on `mapValues(map)` with IN operator. Shell-like completion support in clickhouse keeper-client. Prevent Keeper `mntr` command from getting stuck. Reduce lock contention in Keeper dispatcher. Tolerate missing padding at the end of parquet file blocks. Fix how Alias table targets are saved as DDL dependencies. Fix wrong result or exception during reading subcolumns of ALIAS columns. Fix missing column in JOIN with non-standard identifier alias. Fix crashes in Kusto dialect functions with empty arguments. Forbid mounting local_object_storage outside user_files_path in clickhouse-client. Fix logical race in DeltaLake table engine on snapshot version change. Fix logical error on attaching a part in MergeTree with chained renames. Fix explicit settings being silently ignored when sent with `compatibility`. Fix client reporting NETWORK_ERROR instead of actual parsing error in INSERT with parallel parsing.
</Release>

<Release url="https://clickhouse.com/docs/whats-new/changelog#clickhouse-performance-and-feature-improvements">
## ClickHouse Performance and Feature Improvements

**Performance Improvements:**
- Allow read-in-order optimization and primary-key pruning with `Nullable` CAST target types for monotonic conversions
- Allow index pruning and filter pushdown when comparing integral columns with float literals
- Added SLRU cache for Parquet metadata to improve read performance
- Support swapping sides of ANTI, SEMI and FULL joins based on optimizer statistics
- Optimized granules skipping for `pointInPolygon` and fixed index analysis issues
- Improved `levenshteinDistance` function performance
- Optimized batch decimal type conversions by avoiding per-element function calls
- Iceberg tables now support asynchronous metadata prefetching and cached metadata usage
- S3Queue ordered mode uses ListObjectsV2 StartAfter to reduce ListObjects calls
- Lowered memory usage for inserts deduplication in sync mode
- Use arch-specific cache line size instead of hardcoded 64-byte value
- Optimized text index dictionary reading and analysis
- Sped up LZ4 decompression of 16 byte blocks in ARM
- Refactored tokenization to high-performance interface with SIMD support
- Improved text index analysis for queries with combined conditions
- Improved performance of queries with constant expressions generating large arrays/maps
- Fixed key condition analysis for `DateTime64` primary keys compared with integer constants
- Setting `optimize_syntax_fuse_functions` enabled by default
- Optimized `avgWeighted` aggregate function with local accumulators (~27% improvement for Nullable inputs)
- Improved performance and reduced memory usage for parallel window functions and `arrayFold` workloads
- Improved sorted merges performance
- Optimized `INTERSECT ALL` and `EXCEPT ALL`
- Added `read_in_order_use_virtual_row` optimization support for reverse-order reads
- Reduced cache contention in RIGHT and FULL JOINs
- Optimized `PrefetchingHelper::calcPrefetchLookAhead` with integer arithmetic
- Reduced Keeper memory consumption by replacing `absl::flat_hash_set` with `CompactChildrenSet` (KeeperMemNode reduced from 144 to 128 bytes)

**Feature Improvements:**
- Aggregate projections now correctly supported in views
- Support OUTER to INNER join conversion optimization with `join_use_nulls`
- Improved subcolumns reading with correct sizes calculation
- Separate jemalloc arenas for mark, uncompressed and page caches to avoid memory fragmentation
- Tables with DELETE TTL rules can now use vertical merge algorithm
- Apply data skipping indexes during distributed index analysis
- Secondary index marks prewarmed when `prewarm_mark_cache` setting enabled
- Reduced locking during access control
- Compound AND conditions in row policies and PREWHERE now decomposed for sorting-key atoms extraction
- Reduced lock contention in MergeTreeBackgroundExecutor
- Fixed excessive memory usage (~514 MiB) during format auto-detection for non-Arrow data
- Parse GeoParquet files with different Geo types in same column
- Introduced `tokensForLikePattern` SQL function for LIKE pattern tokenization
- Added `{_schema_hash}` placeholder for S3 table engine
- SymbolIndex, addressToSymbol, system.symbols, buildId now work on macOS
- `system.stack_trace` table now works on macOS
- Added per-server LDAP config option `<follow_referrals>` to control referral chasing
- Track data skipping indices used in query execution via `skip_indices` column in query_log
- ACCESS_DENIED hints no longer reveal column names unless user can show all required columns
- Added dedicated cleanup thread for MergeTree to prevent cleanup delays
- Reload cluster config if IPs of local server's hostname changed
- Allow `optimize_aggregators_of_group_by_keys` to correctly optimize in GROUPING SETS queries
- Keeper-bench: report errors in metrics and generate JSON metrics file
- Added ROLE clause to CREATE USER
- Internal_replication settings can now be set for Replicated database clusters
- New setting `allow_nullable_tuple_in_extracted_subcolumns` controls Tuple subcolumns behavior
</Release>
