MatrixOne-v3.0.10

Release date: Apr 22 2026
MatrixOne version: v3.0.10

MatrixOne 3.0.10 introduces improvements and bug fixes that enhance usability. Below are the major updates.

Key Improvements

Data branch

*Drop database robustness: tolerate dangling hidden-index metadata during DropDatabase so drops don’t fail on stale catalog state (#24019).
*Correctness after GC + restart: fix cases where data-branch diff/merge could return wrong results on 3.0-dev after GC and restart (#24015).
*Diff usability & fidelity: add COLUMNS projection to data-branch diff (#24116); avoid duplicate rows during diff replay (#24141); *preserve the task store during refresh flows (#24145).
*New statement (3.0-dev): add DATA BRANCH PICK to cherry-pick rows by primary key (#24082).

SQL tasks & full-text search

*SQL task pipeline: implement an end-to-end SQL task execution pipeline (#24123); fix body quoting/escaping and stabilize the related test cases (#24125).
*Full-text (3.0-dev cherry-pick): bring in single-keyword fast path and related correctness fixes from #24027 (#24102).

Role model & backports

*Authorization: add a role rule capability (#23823) (#23851).
*Backports to 3.0-dev: cherry-pick #23961 and #23963 (#24028).
*Global stats stability: backport GlobalStats hang/race fixes to 3.0-dev (#24143).

Statistics, subscriptions, and operability

*Statistics view upgrade (3.0): repair the statistics view upgrade path (#24080).
*Subscriptions (3.0): fix cloning of hidden views originating from subscriptions (#24124).
*Account introspection: optimize SHOW ACCOUNTS account-info SQL execution in the frontend layer (#24092).

Key Bug Fixes

Execution engine, types, and TAE

*NULL / varlen handling: avoid a top-level panic on null varlen copy (#24049).
*Decimal comparisons: prevent panics in valueDec128Compare when comparing DECIMAL to integer types (#24088).
*Storage correctness: after TransferDeleteRows, refresh the rowid slice to prevent a dangling pointer in TAE paths (#24096).

Transactions, CDC, and caching

*Transaction lifecycle: fix explicit transaction long-holder leaks on backend failure paths (#24105).
*CDC (3.0): fix memory leaks in the sinker pipeline (#24128).
*Cache correctness: move postEvict callbacks outside queueLock in the FIFO cache to avoid lock-scope hazards (#24117).

Proxy / DistTAE and sessions

*Resource leaks & deadlocks: clean up leaked backend sessions and avoid WriteBatch self-deadlock (#24152).
*Disconnect behavior: avoid cleanup storms on normal client disconnects (#24157).

What's Changed

• fix: tolerate dangling hidden index metadata in DropDatabase by @gouhongshen in #24019
• to 3.0: fix: data branch diff/merge returns wrong results after GC+restart on 3.0-dev by @gouhongshen in #24015
• feat: Add rule for role (#23823) by @ouyuanning in #23851
• to 3.0-dev: cherry-pick #23961 and #23963 by @ck89119 in #24028
• fix: repair statistics view upgrade on 3.0 by @LeftHandCold in #24080
• fix: avoid top panic on null varlen copy by @LeftHandCold in #24049
• fix: prevent panic in valueDec128Compare when comparing decimal with integer types by @jiangxinmeng1 in #24088
• [3.0-dev] feat: add DATA BRANCH PICK statement for cherry-picking rows by PK by @gouhongshen in #24082
• fix explicit txn long-holder leak on backend failures by @LeftHandCold in #24105
• fix(tae): refresh rowids slice after TransferDeleteRows to prevent dangling pointer by @jiangxinmeng1 in #24096
• frontend: optimize SHOW ACCOUNTS account-info SQL by @LeftHandCold in #24092
• feat: implement sql task execution pipeline by @iamlinjunhong in #24123
• 3.0: fix: clone hidden views from subscriptions by @gouhongshen in #24124
• 3.0: cdc: fix memory leaks in sinker pipeline by @jiangxinmeng1 in #24128
• to 3.0: data branch diff: add COLUMNS projection by @gouhongshen in #24116
• fix sql task body quoting and stabilize sql task case by @iamlinjunhong in #24125
• fix: move postEvict callbacks outside queueLock in FIFO cache by @aptend in #24117
• fix: backport GlobalStats hang/race fixes to 3.0-dev by @ck89119 in #24143
• to 3.0: fix: preserve task store during refresh by @gouhongshen in #24145
• to 3.0-dev: cherry-pick #24027 (fulltext single-keyword fast path and correctness fixes) by @XuPeng-SH in #24102
• to 3.0: fix: avoid duplicate rows in data branch diff replay by @gouhongshen in #24141
• proxy/disttae: clean leaked backend sessions and avoid WriteBatch self-deadlock by @LeftHandCold in #24152
• proxy: avoid cleanup storms on normal disconnect by @LeftHandCold in #24157

Full Changelog: v3.0.9...v3.0.10

MatrixOne-v3.0.9

Release date: Apr 1 2026
MatrixOne version: v3.0.9

MatrixOne 3.0.9 introduces improvements and bug fixes that enhance usability. Below are the major updates.

Key Improvements

Vector Search & AI Features

*IVF-FLAT build parity: Port IVF-FLAT index build from main for consistent behavior across branches (#23737).
*IVF create plan: Remove cross apply from ivf_create on 3.0-dev to simplify the plan (#23780).
*IVF runtime robustness: Harden InMem IVF paths against distVec issues in multi-batch reads and persisted data (#23783, #23812).
*IVF memory: Move IVFFLAT-related allocations to heap to avoid stack / escape issues (#23837).
*Vector parsing: Restore dimension validation in strToArray for safer string-to-vector ingestion (#23878).

Lock Service & Transactions

*Lock mode on promotion: When a waiter becomes a holder, keep lock mode in sync; fix range lock mode not synced on the paired entry (#23787, #23809).
*Snapshot / ordering experiments: Revert refreshSnapshotAfterLock (#23794); a clone+drop race fix was merged then reverted *before the final pairing fix (#23799, #23802).
*Lock retry lifecycle: Exit lock retry cleanly when context is done (#23907).
*Message / apply cleanup: Harden cleanup for abandoned JoinMap and nil table functions (#23932).
*Parallel executor: Fix dirty lockcount under concurrency (#23928).

Data Branch, Snapshot & Storage (3.0)

*CI / quality: Fix flaky TestIssue19551 caused by race conditions (#23773).
*Stale read after merge + GC: Correct immediate branch merge interactions after GC (#23827).
*Data branch diff: Improve diff correctness, memory control, and summary output (#23811).
*Full-text + snapshot: Harden fulltext snapshot restore (#23867).
*PK / tombstones: Make tombstone checks PK-aware in PKPersistedBetween on 3.0 (#23869).
*Object I/O metadata: Keep zonemap for hidden rowid and ts columns after refactor (#23850).
*DDL / conflicts: Mitigate intermittent ww_conflict / duplicate key after drop primary key on 3.0 (#23890).
*Multi-CN retry: Detect duplicate keys during multi-CN retry paths (#23917).
*Dedup / snapshot: In dedupjoin, return ErrTxnNeedRetry from checkSnapshotAdvancedDuplicates to avoid false-positive duplicate errors (#23952).

SQL,** Plan & Execution

*Prepared statements: Fix PREPARE with ON DUPLICATE KEY UPDATE when NULL is written to an UNSIGNED column (#23888).
*ODKU semantics: Fix VALUES() returning stale values instead of NULL in multi-row INSERT … ON DUPLICATE KEY UPDATE (#23896).
*Vectors / JSON: Reset vector class to FLAT in ResetWithSameType to fix prepared statement + NULL JSON panic (3.0-dev) (#23865).
*Foreign keys: Support repeated delete plan merges for FK cascade on 3.0 (#23900).
*Windows & HAVING: Support HAVING filters on window aliases (#23951); fix window + ORDER BY on non-numeric types with RANGE UNBOUNDED frame (#23945).
*Dedup join + prepare: Fix prepare interaction with dedup join (#23926).
*Information schema: INDEX_TYPE in statistics should use idx.algo instead of NULL (#23933).

Cluster & Infrastructure

*HAKeeper client: Retry EOF errors for resilience (#23903).
*CDC: Address flaky CDC tests / behavior (#23908).
*Scans: Make protectedScans suspend/resume defer-safe (#23920).
*Object storage: GC / fileservice compatibility fixes for OBS-like S3 backends (#23922).

Key Bug Fixes

Critical Panics & Crashes

*IVF-FLAT InMem paths: distVec / NULL vector scan panics (#23783, #23812, #23835).
*Prepared statement path: NULL JSON + vector reset panic (#23865).
*Data Consistency & Concurrency
*Lock-service waiter/holder and range lock pairing (#23787, #23809).
*Stale read after branch merge and GC (#23827).
*Tombstone vs PK in persisted paths (#23869).
*False duplicates / retry semantics in dedupjoin and multi-CN (#23952, #23917).

CDC & Replication

*Flaky CDC stabilization (#23908).

###SQL & Query Fixes
*ON DUPLICATE KEY UPDATE with UNSIGNED and VALUES() correctness (#23888, #23896).
*Window frames and HAVING on window expressions (#23951, #23945).
*Information_schema INDEX_TYPE (#23933).

Infrastructure & Compatibility

*HAKeeper EOF retries (#23903).
*S3 / OBS-like backend quirks in GC (#23922).
*protectedScans and defer safety (#23920).
*JoinMap / pipeline cleanup (#23932).

What's Changed

• ivfflat index build port from main by @cpegeric in #23737
• to 3.0: fix: resolve TestIssue19551 flaky test due to race conditions by @gouhongshen in #23773
• fix: prevent distVec panic in multi-batch IVFFLAT InMem reads by @iamlinjunhong in #23783
• fix(lockservice): sync lock mode when waiter is promoted to holder by @LeftHandCold in #23787
• revert refreshSnapshotAfterLock by @LeftHandCold in #23794
• remove cross apply from ivf_create 3.0-dev by @cpegeric in #23780
• lockservice,txn: fix clone+drop race by syncing range lock mode and reordering doWrite defers by @LeftHandCold in #23799
• Revert "lockservice,txn: fix clone+drop race by syncing range lock mode and reordering doWrite defers (#23799)" by @LeftHandCold in #23802
• lockservice: fix range lock mode not synced to paired entry on waiter promotion by @LeftHandCold in #23809
• fix persisted without deal with distVec by @iamlinjunhong in #23812
• To 3.0 by @gouhongshen in #23819
• to 3.0: fix: handle stale read on immediate branch merge after GC by @gouhongshen in #23827
• fix: panic on IVF-Flat entries table scan with NULL vectors in InMem … by @iamlinjunhong in #23835
• to 3.0: fix(data-branch): improve diff correctness, memory control, and output summary support by @gouhongshen in #23811
• ivfflat memory escape to heap (3.0-dev) by @cpegeric in #23837
• to 3.0: fix: harden fulltext snapshot restore by @gouhongshen in #23867
• fix: make tombstone check PK-aware in PKPersistedBetween 3.0 by @aptend in #23869
• fix: reset vector class to FLAT in ResetWithSameType to fix prepared stmt null json panic (3.0-dev) by @ck89119 in #23865
• to 3.0: refactor(objectio): keep zonemap for hidden rowid and ts columns by @gouhongshen in #23850
• fix: restore vector dimension validation in strToArray for … by @iamlinjunhong in #23878
• Fix PREPARE with ON DUPLICATE KEY UPDATE for NULL on UNSIGNED column by @jiangxinmeng1 in #23888
• fix: support repeated delete plan merges for fk cascade 3.0 by @aptend in #23900
• fix: retry EOF errors from HAKeeper client by @robll-v1 in #23903
• to 3.0: fix: resolve intermittent ww_conflict duplicate key after drop primary key on 3.0 by @gouhongshen in #23890
• Fix lock retry exit on done context by @LeftHandCold in #23907
• fix: flaky cdc by @XuPeng-SH in #23908
• fix VALUES() returning stale data instead of NULL in multi-row INSERT ON DUPLICATE KEY UPDATE by @jiangxinmeng1 in #23896
• fix: catch duplicate keys during multi-CN retry by @gouhongshen in #23917
• fileservice/gc: fix S3 compatibility issues for OBS-like backends by @LeftHandCold in #23922
• fix: make suspend/resume of protectedScans defer-safe by @XuPeng-SH in #23920
• fix: prepare and dedup join by @XuPeng-SH in #23926
• fix: dirty lockcount by @XuPeng-SH in #23928
• fix(message,apply): harden cleanup for abandoned JoinMap and nil table functions by @LeftHandCold in #23932
• fix(information_schema): statistics INDEX_TYPE should use idx.algo instead of NULL by @ck89119 in #23933
• fix(plan): support HAVING filters on window aliases by @iamlinjunhong in #23951
• to 3.0: fix(dedupjoin): return ErrTxnNeedRetry in checkSnapshotAdvancedDuplicates to fix false-positive duplicate entry (#23943) by @gouhongshen in #23952
• fix: window function with ORDER BY on non-numeric type errors for RANGE UNBOUNDED frame by @ck89119 in #23945

Full Changelog: v3.0.8...v3.0.9

MatrixOne-v3.0.8

Release date: Feb 27 2026
MatrixOne version: v3.0.8

MatrixOne 3.0.8 introduces bug fixes that enhance usability. Below are the major updates.

Bug Fixes

1. Fix refresh snapshot after exclusive lock to prevent orphan records from concurrent CLONE
2. Fix prebuild composite-unique lock column mapping for INSERT

What's Changed

• to 3.0: fix(sql/plan): prebuild composite-unique lock column mapping for INSERT by @gouhongshen in #23763
• [Fix] DropDatabase: refresh snapshot after exclusive lock to prevent orphan records from concurrent CLONE by @LeftHandCold in #23767

Full Changelog: v3.0.7...v3.0.8

MatrixOne-v3.0.7-hotfix-20260223

Release date: Feb 23 2026
MatrixOne version: v3.0.7-hotfix-20260223

MatrixOne 3.0.7-hotfix-20260223 introduces bug fixes that enhance usability. Below are the major updates.

Bug Fixes

Fix build failed on arm env

What's Changed

• BUG: check valid haswell before substitute by @cpegeric in #23756

Full Changelog: v3.0.7...v3.0.7-hotfix-20260223

MatrixOne-v3.0.7

Release date: Feb 22 2026
MatrixOne version: v3.0.7

MatrixOne 3.0.7 introduces bug fixes that enhance usability. Below are the major updates.

Bug Fixes

1.Fix pipeline client leak
2.Fix execute create as select in one transaction and enforce source SELECT privilege
3.Fix force CTAS follow-up insert to run on one CN
4.Fix UNION with ORDER BY FIELD() function
5.Fix cdc truncate bug

What's Changed

• fix cdc truncate by @jiangxinmeng1 in #23623
• Fix UNION with ORDER BY FIELD() function by @aptend in #23627
• disable wrong parallel decompression for LOAD DATA by @robll-v1 in #23648
• fix: fix pipeline client leak 3.0-dev by @XuPeng-SH in #23711
• to 3.0: fix: UT TestPauseResumeDaemonTask deadwait by @gouhongshen in #23734
• to 3.0: fix: execute create as select in one transaction and enforce source SELECT privilege by @gouhongshen in #23726
• to 3.0: fix: force CTAS follow-up insert to run on one CN by @gouhongshen in #23749

Full Changelog: v3.0.6...v3.0.7

MatrixOne-v3.0.6

Release date: Jan 28 2026
MatrixOne version: v3.0.6

MatrixOne 3.0.6 introduces improvements and bug fixes that enhance usability. Below are the major updates.

Key Improvements

1. Bloom filter & multi‑CN: Support Bloom filters without regular filters and broadcast Bloom filters across CNs to improve distributed query performance.
2. Backup & export: Add a file index for backups and fix table data export issues to make backup/restore and export more reliable.
3. Vector features: Enable simsimd (MO_CL_SIMSIMD=1), support vector indexes for prepared statements, and refine vector test cases for better performance and stability.
4. Optimizer & execution: Improve UNION result stability, refine filter push‑down and hint handling (including CTE and type matching), and correct NDV estimation for modulo operations.
5. GC & storage: Optimize GC behavior, fix tmp‑fs GC logging nil‑entry handling, and scope CN segment maps by transaction and table to avoid clone‑related delete skips.

Bug Fixes

1. Fix a hang that could occur after packet loss during data loading.
2. Fix inconsistent caching in the auto‑increment service, stabilize nested SQL/auto‑increment retry paths, and remove noisy auto‑increment logs.
3. Fix multi‑CN scan statistics consistency, a window function compatibility issue, and cap out of range crashes.
4. Correct default MARCH for x86_64 (was incorrectly set to haswell) to avoid compatibility and performance issues.

What's Changed

• modify year from 2025 to 2026. by @Ariznawlll in #23441
• use bf without filter and broadcast bf with multi cn by @iamlinjunhong in #23446
• alter cases by @heni02 in #23466
• Add file index for backup by @LeftHandCold in #23474
• Fixed a bug that caused the system to stuck after packet loss during data loading by @LeftHandCold in #23390
• remove a useless warning 3.0 by @aptend in #23482
• Fixed a bug related to inconsistent caching in the auto-increment service by @LeftHandCold in #23489
• to 3.0: Fix multi-CN scan collection consistency and document AP shuffle/merg… by @gouhongshen in #23492
• enable simsimd library with MO_CL_SIMSIMD=1 by @cpegeric in #23490
• modify vector case for stable test by @iamlinjunhong in #23496
• fix: Correct NDV estimation for modulo operations to 3.0-dev by @XuPeng-SH in #23486
• chore: change code owner 3.0-dev by @XuPeng-SH in #23500
• to 3.0: Fix nil entry handling in tmp fs GC logging by @gouhongshen in #23479
• order for union for stable test by @iamlinjunhong in #23505
• to 3.0: Stabilize retry paths for nested SQL and auto‑increment allocation by @gouhongshen in #23343
• enable vector index for prepared stmt by @aunjgr in #23517
• Fix window function compatibility bug by @LeftHandCold in #23515
• to 3.0: fix export table data by @gouhongshen in #23539
• fix remap hint after cte and match int to float by @iamlinjunhong in #23537
• GC optimization by @LeftHandCold in #23491
• BUG: default MARCH set to haswell for x86_64 by @cpegeric in #23547
• to 3.0: Scope CN segment map by txn and table to prevent clone-induced delete skips by @gouhongshen in #23556
• #23579 fix cap out of range by @cpegeric in #23585
• remove auto-increment log by @LeftHandCold in #23596
• to 3.0: unlimited filtering on workspace entry by @gouhongshen in #23595

Full Changelog: v3.0.5...v3.0.6

MatrixOne-v3.0.5

Release date: Jan 5 2026
MatrixOne version: v3.0.5

MatrixOne 3.0.5 introduces improvements and bug fixes that enhance usability. Below are the major updates.

Key Improvements

Vector Search & AI Features

• Vector Index Enhancements: Removed experimental flags for IVF and FullText indexes, making them production-ready
• Vector Pre-filtering: Added support for enable_vector_prefilter_by_default variable and IVF pre-filtering
• Vector Distance Output: Direct vector distance output by Reader for better performance
• Normalized L2 Support: Allow normalized_l2 to use vector index

Data Branch & Git4Data

• Data Branch Operations: Full support for data branch operations including create, diff, and stage
• Database/Table Management: Support for database/table deletion in data branch
• Parallel Processing: Data branch diff parallel writing to remote storage
• Context-aware Keywords: Make data-branch keywords context-aware

CDC (Change Data Capture)

• CDC Reimplementation: Complete CDC reimplement based on existing design
• GC Support: GC now supports CDC functionality
• Watermark Management: Improved CDC watermark handling and cleanup
• Metric Dashboard: Added CDC metric dashboard for monitoring

Performance & Optimization

• Index Selection: Optimized index selection algorithms
• LIMIT BY RANK: Performance optimization for LIMIT BY RANK queries
• FullText Pushdown: FullText pushdown limit when no filter
• Query Metrics: Added ReadSize, S3ReadSize, and DiskReadSize metrics to EXPLAIN ANALYZE

Infrastructure & Development

• Role Management: Support for rename role name functionality
• Multi-CN Development: Added multi-cn docker compose file for better development
• Go Version: Upgraded Go version to 1.25.4
• Logging Optimization: Removed/optimized logging to reduce noise

Key Bug Fixes

Critical Panics & Crashes

• Multi-CN IVF Search: Fixed panic during multi-cn ivf search
• JOIN Condition Conversion: Fixed panic when JOIN condition type conversion fails
• COUNT DISTINCT: Fixed panic when using count(distinct(, ))
• ORDER BY in JOIN: Fixed ORDER BY panic in JOIN queries
• Remote Workspace: Fixed remote run workspace panic
• Insert Ignore: Fixed insert ignore panic

Data Consistency Issues

• Multi-update Null Mask: Fixed null mask mutation during shrink in multi_update
• Data Inconsistency: Fixed data inconsistency due to stale lock service flags
• Index Table Consistency: Fixed index table data inconsistency in INSERT ON DUPLICATE KEY UPDATE
• Stale Read: Multiple fixes for stale read issues
• Batch Reuse: Fixed data mapping error when batch reused with different Vecs/Attrs

CDC-specific Fixes

• CDC Data Loss: Fixed CDC pause/resume/restart data loss
• CDC Restart: Fixed downstream "delete" loss after CDC upstream restart
• CDC Watermark: Fixed CDC watermark updater and stale watermark cleanup
• CDC Stale Read: Fixed CDC stale read issues

SQL & Query Fixes

• Foreign Key Actions: Correctly format foreign key actions in ConstructCreateTableSQL
• Column Pruning: Fixed column pruning failure when applying associative law rules
• JOIN Remap: Fixed remap error in JOIN with cross-table column references
• Left JOIN OR: Fixed left join OR predicate regression
• Affected Rows: Fixed delete/replace statement reporting incorrect affected rows

Type Conversion & Casting

• TEXT to CHAR/VARCHAR: Skip length validation when casting TEXT to CHAR/VARCHAR
• TIMESTAMP Parsing: LOAD DATA now uses session timezone for TIMESTAMP parsing
• Fractional Precision: Enforce fractional-second precision end-to-end for TIME/DATETIME

Infrastructure Fixes

• Autocommit Sync: Fixed autocommit status sync issue
• Connection Balancing: Fixed proxy connection balancing when connecting too fast
• Client Pool: Fixed client pool random CI errors
• Version Information: Fixed lost version information when building images

What's Changed

• (pick to 3.0-dev): fix slience error by @XuPeng-SH in #22769
• (pick to 3.0-dev): add multi-cn docker compose file for better development by @XuPeng-SH in #22766
• fix insert ignore panic 3.0 by @aptend in #22755
• (pick to 3.0-dev): fix decimal stats bug by @XuPeng-SH in #22767
• parquet support int8 int16 binary by @jiangxinmeng1 in #22771
• (pick to 3.0-dev): IF function with GROUPING in ROLLUP by @XuPeng-SH in #22776
• allow normalized_l2 to use vector index by @aunjgr in #22788
• fix change handle by @jiangxinmeng1 in #22786
• (3.0): reimplement cdc base on the existing design by @XuPeng-SH in #22801
• Fix index table data inconsistency in INSERT ON DUPLICATE KEY UPDATE 3.0 by @aptend in #22782
• (metric): add cdc metric dashboard for 3.0-dev by @XuPeng-SH in #22810
• fix cdc stale read by @jiangxinmeng1 in #22807
• Set the maximum number of load retries and retry time by @LeftHandCold in #22812
• (fix): parallize open logservice clients by @XuPeng-SH in #22804
• (fix): cdc stale watermark cleanup 3.0 by @XuPeng-SH in #22815
• sync commit ts in cdc create task by @jiangxinmeng1 in #22770
• (fix): left join OR predicate regression 3.0-dev by @XuPeng-SH in #22817
• (fix): Fix remap error in JOIN with cross-table column references 3.0 by @aptend in #22826
• update cdc user guide by @jiangxinmeng1 in #22823
• fix stale read by @jiangxinmeng1 in #22839
• fix: Split large sels into multiple batches in CompactBatchs.Union 3.0 by @aptend in #22843
• Enforce fractional-second precision end-to-end for TIME/DATETIME/TIME… by @XuPeng-SH in #22728
• fulltext pushdown limit when no filter 3.0-dev by @cpegeric in #22848
• fix: the proxy cannot balance connections when connect too fast by @volgariver6 in #22856
• (fix): client pool random ci error 3.0-dev by @XuPeng-SH in #22862
• to 3.0: adding config for the prefetching on the subscribed tables by @gouhongshen in #22868
• feat: Add ReadSize, S3ReadSize, and DiskReadSize metrics to EXPLAIN ANALYZE 3.0-dev by @XuPeng-SH in #22867
• bug fix stream block when user abort by @cpegeric in #22917
• fix: distinct agg should be run in only one node and without any parallel 3.0-dev by @XuPeng-SH in #22923
• Supports the dayofweek function by @LeftHandCold in #22922
• cherry-pick vector-index related PRs to 3.0-dev by @aunjgr in #23053
• Fix: LOAD DATA should use session timezone for TIMESTAMP parsing 3.0-dev by @XuPeng-SH in #22928
• FEATURE optimize for LIMIT BY RANK by @cpegeric in #23059
• (fix): cdc watermark updater by @XuPeng-SH in #22821
• changes handle skip deletes by @jiangxinmeng1 in #23114
• fix: ORDER BY panic in JOIN queries by @XuPeng-SH in #23141
• rm tag 19998(3.0-dev) by @Ariznawlll in #22759
• GC support CDC by @LeftHandCold in #22779
• fix lost version information when build image (#22830) by @guguducken in #23082
• CI: upgrade go version to 1.25.4 for 3.0-dev by @guguducken in #23151
• Fixed an issue where the StopSending message was incorrectly used to cancel a dispatch receiver by @LeftHandCold in #23159
• pre filter for ivf by @iamlinjunhong in #23117
• output vector distance directly by Reader by @aunjgr in #23171
• (fix): data mapping error when batch reused with different Vecs/Attrs - 3.0-dev by @XuPeng-SH in #23166
• to 3.0: supporting data branch operations by @gouhongshen in #23152
• feat: Remove experimental flags for IVF and FullText indexes 3.0-dev by @XuPeng-SH in #23174
• fix: skip length validation when casting TEXT to CHAR/VARCHAR by @zhangxu19830126 in #23178
• GC supports backup protection for 3.0 by @LeftHandCold in https://github.com/matrixorigin/matrixone/...

MatrixOne-v3.0.4-hotfix-20251226

Release date: Dec 26 2025
MatrixOne version: v3.0.4-hotfix-20251226

MatrixOne 3.0.4-hotfix-20251226 introduces bug fixes that enhance usability. Below are the major updates.

Bugfix

Fix snapshot bug
skip length validation when casting TEXT to CHAR/VARCHAR

What's Changed

• [fix] update dragonboat version, fix the snapshot bug: by @volgariver6 in #23281
• skip length validation when casting TEXT to CHAR/VARCHAR to hotfix by @zhangxu19830126 in #23370

Full Changelog: v3.0.4-hotfix-20251213...v3.0.4-hotfix-20251226

MatrixOne-v3.0.4-hotfix-20251213

Release date: Dec 13 2025
MatrixOne version: v3.0.4-hotfix-20251213

MatrixOne 3.0.4-hotfix-20251213 introduces bug fixes that enhance usability. Below are the major updates.

Bugfix

Fix clone indexes when alter table

What's Changed

• to hotfix: fix clone indexes when alter table. by @gouhongshen in #23260

Full Changelog: v3.0.4-hotfix-20251210...v3.0.4-hotfix-20251213

MatrixOne-v3.0.4-hotfix-20251210

Release date: Dec 10 2025
MatrixOne version: v3.0.4-hotfix-20251210

MatrixOne 3.0.4-hotfix-20251210 introduces bug fixes that enhance usability. Below are the major updates.

Bugfix

Fix block filter

What's Changed

• to hotfix: fix block filter by @gouhongshen in #23239

Full Changelog: v3.0.4-hotfix-20251204...v3.0.4-hotfix-20251210