docs: update roadmap

docs/en/29-roadmap/index.md

@@ -6,9 +6,9 @@ The 2025 roadmap for TDengine OSS is described in the following table.
 
 |  Quarter   |  Feature  |
 | :----- | :----- |
-| 2025Q1 | <ol><li>Virtual tables</li><li>Query engine: conditional expressions in <code>REGEXP</code>, <code>GREATEST</code>, <code>LEAST</code>, and <code>CAST</code> functions; improvements in single-row selection functions; time range interpolation with <code>INTERP</code></li><li>Storage engine: support for writing query results into supertables; <code>KEEP</code> parameter for supertables; performance improvements for the parameter binding interface</li><li>Stream processing: support for virtual tables; decreased compute resource usage; new mechanism for event notification; faster stream creation</li><li>Data types: support for the decimal data type</li><li>High availability: faster recovery from downtime; improved client failover</li><li>Stability: LTS release TDengine 3.3.6.x</li><li>JDBC driver: more efficient data ingestion</li><li>Ecosystem: integration with Microsoft Excel</li></ol> |
-| 2025Q2 | <ol><li>Query engine: relaxed restrictions on <code>JOIN</code> queries; support for all mathematical functions in MySQL; integral, integral average, and continuous variance functions; optimization of the <code>CSUM</code> function; support for <code>COUNT(DISTINCT)</code> syntax; enhancements to event windows; faster filtering by tag; faster <code>INTERP</code> queries</li><li>Storage engine: decreased compute resource usage for TSMAs; improved write jitter</li><li>Stream processing: high availability of snodes</li><li>Data types: support for the blob data type</li><li>Data subscription: support for the MQTT protocol</li><li>High availability: faster replica configuration changes; faster recovery from downtime for clusters; improved data recovery after power outage</li><li>Observability: diagnostic tool for data ingestion</li></ol> |
-| 2025Q3 | <ol><li>Query engine: more subqueries; support for all operators in MySQL; support for all time functions in MySQL; improved window calculation; reduced jitter in query performance; support for specifying columns in count windows</li><li>Storage engine: faster ingestion in SQL mode</li><li>Observability: diagnostic tool for queries; improved <code>EXPLAIN</code> output; monitoring of long-running tasks</li></ol> |
-| 2025Q4 | <ol><li>Query engine: window functions (i.e. the <code>OVER</code> clause); support for all string, aggregation, and conditional functions in MySQL; sorting within groups for partition queries; controls for query resource usage; faster aggregate queries on subtables; time range interpolation in <code>INTERVAL</code> windows</li><li>Data types: support for variable-length strings</li><li>Caching: faster row-oriented caching</li><li>Observability: more insight into operations and maintenance</li></ol> |
+| 2025Q1 | <ol><li>​​Virtual Tables​​</li><li>​​Query Capabilities​​: Support using constant values as JOIN primary keys; support for REGEXP, GREATEST, LEASTfunctions; support for conditional expressions in the CASTfunction; support for retrieving other column values from the row returned by a single-row selection function via the colsfunction; INTERPsupports interpolation time ranges; performance improvements for down-sampling and GROUP BY tag queries.</li><li>​​Storage Capabilities​​: Support controlling the retention period of supertable time-series data via the KEEPparameter; STMT write performance improvements; support for reorganizing tdbfiles; support for dynamic adjustment of the dataDirparameter in multi-tiered storage; improved write performance for automatic table creation.</li><li>​​Stream Processing​​: Support using virtual tables in stream processing; support for window event notification mechanism; support for setting minimum duration limits for event windows and state windows; support for continue_window_closetrigger mode; computational resource optimization; creation time optimization.</li><li>​​Data Types​​: DECIMAL</li><li>​​High Availability​​: Accelerated downtime recovery; optimized client Failovermechanism; significantly improved data synchronization performance between clusters; support for viewing Vnoderecovery progress via SQL.</li><li>​​Stability​​: Begin maintenance of the new stable version 3.3.6.x</li><li>​​TDgpt​​: Support integration with the Community Edition.</li><li>​​JDBC​​: Efficient writing.</li><li>​​Ecosystem Tools​​: Integration with Tableau; integration with Excel.</li></ol> |
+| 2025Q2 | <ol><li>​​Storage Capabilities​​: Multi-tiered storage supports shared storage; support for data mounting; support for writing query results to supertables; support for encrypting user passwords using the SM4algorithm; optimization of STMT2behavior.</li><li>​​Stream Processing​​: Stream processing refactoring.</li><li>​​Data Subscription​​: Support for MQTTprotocol; optimized data subscription offsetupdate logic.</li><li>​​Data Types​​: BLOB</li><li>​​High Availability​​: Improved performance for node recovery and replica changes; faster cluster downtime recovery; optimized power-off data recovery mechanism.</li><li>​​Observability​​: Support for more write performance monitoring metrics.</li><li>​​Communication​​: Support for IPv6.</li><li>​​TDgpt​​: Support for covariate analysis prediction; support for Amazon's Time Series Foundation Model.</li><li>​​Ecosystem Tools​​: Integration with FanRuan FineBI.</li></ol> |
+| 2025Q3 | <ol><li>​​Query Capabilities​​: Support for configuring state window start/end points; count windows allow specifying columns; PARTITION BY TAGsupports string operations; support for all MySQL aggregation and conditional functions (newly added std, variance, stddev_samp, var_samp, group_concat, if, isnull, isnotnull, nvl, coalesce, nullif, ifnullfunctions); support for find_in_set, regexp_in_set, like_in_setfunctions; support for datefunction; improved concurrency for lastqueries; separation of data query and metadata query processing threads.</li><li>​​Storage Capabilities​​: Support for Rollup SMA; support for Timewise-Range SMA; support for scanning the integrity of time-series data files via the scancommand; refactored state control mechanism for shared storage data migration; added show ssmigratecommand to view shared storage migration progress; shared storage supports NAS; floating-point numbers support BSSencoding.</li><li>​​Stream Processing​​: Computational performance and resource efficiency optimization; support for notification filtering mechanism; support for identifying virtual table changes; support for operations on the statecolumn within state windows; support for delete_output_tableoption.</li><li>​​High Availability​​: Adjusted execution logic of the BALANCE VGROUP LEADERcommand for more even distribution of VGroupLeaders.</li><li>​​TDgpt​​: Support for data completion functionality.</li><li>​​Communication​​: Support for TLScertificate authentication transmission.</li></ol> |
+| 2025Q4 | <ol><li>​​Virtual Tables​​: Support for inheritance mechanism; support for custom columns in child tables; virtual supertable query performance improvement; maximum number of virtual table columns increased to 65535.</li><li>​​Query Capabilities​​: Improved query performance for last_row + tags; improved query performance for last_row + composite key; improved efficiency for event window queries grouped by tbname; support for outputting the actual start/end times of the outermost two windows; fill(prev)supports filling with the previous non-NULLvalue.</li><li>​​Storage Capabilities​​: Support for modifying RSMA.</li><li>​​Stream Processing​​: Support for triggering by natural month; support for sub-event triggering; support for interpand percentilefunctions; further performance and stability improvements; optimized memory usage.</li><li>​​Data Subscription​​: Subscription updates immediately effective after modifying tag values.</li><li>​​Security & Reliability​​: Enhanced capabilities in identity authentication, access control, transmission security, storage security, security auditing, security functions, encryption algorithms, and code security.</li><li>​​TDgpt​​: Support for correlation analysis; support for foundational model fine-tuning; support for Pearson correlation coefficient calculation function.</li></ol> |
 
 For more information, see [TDengine Public Roadmap](https://github.com/orgs/taosdata/projects/4).

docs/zh/29-roadmap/index.md

@@ -6,9 +6,9 @@ TDengine TSDB OSS 之 2025 年年度路线图如下表所示。
 
 |  季度   |  功能  |
 | :----- | :----- |
-| 2025Q1 | <ol><li>虚拟表</li><li>查询能力：<code>REGEXP</code>、<code>GREATEST</code>、<code>LEAST</code>、<code>CAST</code> 函数支持判断表达式、单行选择函数的其他列值、<code>INTERP</code> 支持插值时间范围</li><li>存储能力：支持将查询结果写入超级表、超级表支持 <code>KEEP</code> 参数、STMT 写入性能提升</li><li>流计算：支持虚拟表、计算资源优化、事件通知机制、创建时间优化</li><li>数据类型：Decimal</li><li>高可用：加快宕机恢复时间、优化客户端 Failover 机制</li><li>稳定性：开始维护新的稳定版本 3.3.6.x</li><li>JDBC：高效写入</li><li>生态工具：对接 Tableau</li><li>生态工具：对接 Excel</li></ol> |
-| 2025Q2 | <ol><li>查询能力：大幅放宽关联查询限制、支持 MySQL 所有数学函数、支持积分/积分平均/连续方差函数、<code>CSUM</code> 函数优化、<code>COUNT(DISTINCT)</code> 语法、事件窗口功能增强、提升标签过滤性能、提升 <code>INTERP</code> 查询性能</li><li>存储能力：TSMA 计算资源优化、写入抖动优化</li><li>流计算：节点高可用</li><li>数据类型：BLOB</li><li>数据订阅：支持 MQTT 协议</li><li>高可用：提高副本变更速度、提高集群宕机恢复速度、优化断电数据恢复机制</li><li>可观测性：写入诊断工具</li><li>生态工具：对接帆软 FineBI</li></ol> |
-| 2025Q3 | <ol><li>查询能力：支持更多子查询类型、支持 MySQL 运算符、支持 MySQL 所有时间函数、窗口计算逻辑优化、查询性能抖动、计数窗口允许指定列</li><li>存储能力：提高 SQL 模式写入速度</li><li>可观测性：查询诊断工具、优化 <code>EXPLAIN</code> 输出、长任务观测</li></ol> |
-| 2025Q4 | <ol><li>查询能力：窗口函数（<code>OVER</code> 子句）、支持 MySQL 所有字符串/聚合/条件函数、Partition 支持组内排序、控制查询资源占用、提高子表聚合查询性能、<code>INTERVAL</code> 窗口支持插值时间范围</li><li>数据类型：支持不定长度字符串数据类型</li><li>数据缓存：提升按行缓存性能</li><li>可观测性：增强运维可观测性</li></ol> |
+| 2025Q1 | <ol><li>虚拟表</li><li>查询能力：支持将常量值作为连接主键，支持 <code>REGEXP</code>、<code>GREATEST</code>、<code>LEAST</code> 函数，支持在 <code>CAST</code> 函数中使用判断表达式，支持通过 <code>cols</code> 函数获取单行选择函数返回结果所在行的其他列值，<code>INTERP</code> 支持插值时间范围，提高降采样和按标签分组查询的性能</li><li>存储能力：支持通过 <code>KEEP</code> 参数控制超级表的时序数据保存时长，STMT 写入性能提升，支持重新整理 <code>tdb</code> 文件、支持对多级存储中的 <code>dataDir</code> 参数进行动态调整，提高自动建表的写入性能</li><li>流计算：支持在流计算中使用虚拟表，支持窗口事件通知机制，支持为事件窗口和状态窗口设置最小持续时长限制，支持 <code>continue_window_close</code> 触发模式，计算资源优化，创建时间优化</li><li>数据类型：<code>Decimal</code></li><li>高可用：加快宕机恢复时间，优化客户端 <code>Failover</code> 机制，大幅提升集群之间的数据同步性能，支持通过 <code>SQL</code> 查看 <code>Vnode</code> 恢复进度</li><li>稳定性：开始维护新的稳定版本 <code>3.3.6.x</code></li><li>TDgpt：支持与社区版本对接</li><li>JDBC：高效写入</li><li>生态工具：对接 <code>Tableau</code>，对接 <code>Excel</code></li></ol> |
+| 2025Q2 | <ol><li>存储能力：多级存储支持采用共享存储，支持数据挂载，支持将查询结果写入超级表，支持采用 <code>SM4</code> 算法加密用户密码，优化 <code>STMT2</code> 的行为</li><li>流计算：流计算重构</li><li>数据订阅：支持 <code>MQTT</code> 协议，优化数据订阅 <code>offset</code> 更新逻辑</li><li>数据类型：<code>BLOB</code></li><li>高可用：提升节点恢复和副本变更的性能，提高集群宕机恢复速度，优化断电数据恢复机制</li><li>可观测性：支持更多的写入性能监控指标</li><li>通信：支持 <code>IPv6</code></li><li>TDgpt：支持协变量分析预测，支持亚马逊的时序基础模型</li><li>生态工具：对接帆软 <code>FineBI</code></li></ol> |
+| 2025Q3 | <ol><li>查询能力：支持状态窗口起止点配置，计数窗口允许指定列，<code>partition by tag</code> 支持字符串运算，支持 MySQL 的全部聚合与条件函数（新增 <code>std</code>，<code>variance</code>，<code>stddev_samp</code>，<code>var_samp</code>，<code>group_concat</code>，<code>if</code>，<code>isnull</code>，<code>isnotnull</code>，<code>nvl</code>，<code>collesce</code>，<code>nullif</code>，<code>ifnull</code> 函数，支持 <code>find_inset</code>，<code>regexp_inset</code>，<code>like_in_set</code> 函数，支持 <code>date</code> 函数，提高 <code>last</code> 查询的并发度，分离数据查询和元数据查询处理线程</li><li>存储能力：支持 <code>Rollup SMA</code>，支持 <code>Timewise-Range SMA</code>，支持通过 <code>scan</code> 命令扫描时序数据文件的完整性，改造共享存储数据迁移的状态控制机制，增加 <code>show ssmigrate</code> 命令用于查看共享存储迁移进度，共享存储支持 <code>NAS</code>，浮点数支持 <code>​​BSS</code> 编码</li><li>流计算：计算性能与资源效率优化，支持通知过滤机制，支持识别虚拟表变更，支持在状态窗口中对 <code>state</code> 列进行运算，支持 <code>delete_output_table</code> 选项</li><li>高可用：调整 <code>Balance Vgroup Leader</code> 命令的执行逻辑，使 <code>VGroup</code> 的 <code>Leader</code> 分布更加均匀</li><li>TDgpt：支持数据补全功能</li><li>通信：支持 <code>TLS</code> 证书认证传输</ol> |
+| 2025Q4 | <ol><li>虚拟表：支持继承机制、支持子表自定义列、虚拟超级表查询性能提升，虚拟表列数量上限调整为 <code>65535</code></li><li>查询能力：提升 <code>last_row + tags</code> 的查询性能,提升 <code>last_row + composite key</code> 的查询性能,提升事件窗口按 <code>tbname</code> 分组查询的效率，支持输出最外侧两个窗口的实际起止时间，<code>fill(prev)</code> 支持填充前一个非 <code>NULL</code> 值</li><li>存储能力：支持修改 <code>RSMA</code></li><li>流计算：支持按自然月触发，支持子事件触发，支持 <code>interp</code> 和 <code>percentile</code> 函数，性能和稳定性进一步提升，内存占用优化</li><li>数据订阅：修改标签值后订阅及时生效</li><li>安全可靠性：提升身份鉴别、访问控制、传输安全、存储安全、安全审计、安全函数、加密算法、代码安全方面的能力</li><li>TDgpt：支持相关性分析，支持基础模型微调，支持皮尔逊相关系数计算函数</li>
 
 欲了解更多信息，请参见 [TDengine TSDB Public Roadmap](https://github.com/orgs/taosdata/projects/4) 。