feat: 时间线回退功能 - 代码/文档/测试 by facetosea · Pull Request #35192 · taosdata/TDengine

facetosea · 2026-04-21T09:07:00Z

parser: 无主键时回退到第一个时间戳列
builtins: 放宽 elapsed 等函数主键检查
docs: 中文需求文档(行为对比/函数分类)
test: 54条SQL + 298行期望输出(文件比较模式)

Description

Issue(s)

Close/close/Fix/fix/Resolve/resolve: Issue Link

Checklist

Please check the items in the checklist if applicable.

Is the user manual updated?
Are the test cases passed and automated?
Is there no significant decrease in test coverage?

parser: 无主键时回退到第一个时间戳列 builtins: 放宽 elapsed 等函数主键检查 docs: 中文需求文档(行为对比/函数分类) test: 54条SQL + 298行期望输出(文件比较模式)

gemini-code-assist

Code Review

This pull request introduces the 'Timeline Fallback' feature, which allows timeline and window functions to automatically use the first available TIMESTAMP column when the primary key timestamp is missing. The changes include a new design document, modifications to the timestamp validation logic in builtins.c, and the addition of comprehensive test cases. Review feedback correctly identifies that several core implementation files described in the design document (such as parTranslater.c) are missing from the current patch. Additionally, suggestions were made to rename validation functions to reflect their expanded scope and to replace placeholder Jira IDs in the test documentation.

gemini-code-assist · 2026-04-21T09:08:03Z

+
+## 4. 实现要点
+
+### 4.1 修改点1：`findAndSetTempTableColumn()` (parTranslater.c)


PR 描述和设计文档中详细说明了 parTranslater.c 中的多项核心改动（如 findAndSetTempTableColumn、resetResultTimeline、translateSetOperOrderBy 和 checkSessionWindow），但这些文件并未包含在当前的补丁中。这会导致时间线回退功能在子查询、UNION ALL 和 SESSION 窗口等场景下无法生效。请确认是否遗漏了相关文件的提交。

gemini-code-assist · 2026-04-21T09:08:03Z

 static int32_t checkPrimTS(SNode* pNode, bool* isMatch) {
  int32_t code = TSDB_CODE_SUCCESS;
-  if (nodeType(pNode) != QUERY_NODE_COLUMN || !IS_TIMESTAMP_TYPE(getSDataTypeFromNode(pNode)->type) ||
-      !((SColumnNode*)pNode)->isPrimTs) {
+  if (nodeType(pNode) != QUERY_NODE_COLUMN || !IS_TIMESTAMP_TYPE(getSDataTypeFromNode(pNode)->type)) {


由于该函数现在允许任何时间戳列而不仅仅是主键，函数名 checkPrimTS 和错误码 TSDB_CODE_FUNC_FUNTION_PARA_PRIMTS 变得具有误导性。建议将其重命名为 checkTSColumn 并更新错误码，以符合其当前的逻辑功能，提高代码的可维护性。

gemini-code-assist · 2026-04-21T09:08:03Z

+
+Since:  v3.4.1.0
+Labels: function, timeseries, timeline_fallback, subquery
+Jira:   TS-XXXX


Jira 字段使用了占位符 TS-XXXX，请替换为实际的任务 ID 以完善测试文档。

Copilot

Pull request overview

该 PR 为“时间线回退（Timeline Fallback）”补齐解析器/内置函数参数校验的支持，并新增对应的文档与基于文件对比的 SQL 回归用例，以便在子查询/UNION ALL 等无主键时间列场景下，尽可能回退到首个可用 TIMESTAMP 列建立时间线。

Changes:

parser：临时表找不到 _rowts 时回退到投射列表中的首个 TIMESTAMP 表达式；并放宽 ORDER BY / SESSION 窗口的时间线判定
builtins：放宽 FUNC_PARAM_MUST_BE_PRIMTS 的校验逻辑（允许非主键 TIMESTAMP）
tests/docs：新增中文需求设计文档与覆盖多场景的 SQL 文件对比测试，并接入 CI task

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
test/ci/cases.task	将新增的 timeline fallback 用例纳入 CI 执行列表
test/cases/11-Functions/04-Timeseries/test_fun_ts_timeline_fallback.py	新增基于 `.in/.ans` 文件对比的 pytest 驱动用例
test/cases/11-Functions/04-Timeseries/in/test_timeline_fallback.in	新增 54 条 SQL，覆盖无主键/UNION/窗口/边界与仍需报错场景
test/cases/11-Functions/04-Timeseries/ans/test_timeline_fallback.ans	新增 298 行期望输出（含报错信息）用于回归对比
source/libs/parser/src/parTranslater.c	实现临时表 `_rowts` 回退、ORDER BY 时间线判定放宽、SESSION 列校验放宽
source/libs/function/src/builtins.c	放宽 `checkPrimTS()`：仅要求 TIMESTAMP 列，不再要求 `isPrimTs`
docs/design/timeline-fallback-requirement.md	新增中文需求/行为对比/实现点/测试覆盖说明文档

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-04-21T09:13:03Z

+  if (!isPrimaryKeyImpl((SNode*)pSession->pCol) &&
+      !IS_TIMESTAMP_TYPE(((SColumnNode*)pSession->pCol)->node.resType.type)) {
    PAR_ERR_RET(generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_INTER_SESSION_COL));


checkSessionWindow() now casts pSession->pCol to SColumnNode* to read node.resType.type without first verifying the node is a column. If pCol is ever not QUERY_NODE_COLUMN (e.g., parser produced an expression), this becomes undefined behavior and can crash. Add an explicit nodeType(pSession->pCol) == QUERY_NODE_COLUMN guard (and preferably use getSDataTypeFromNode() / IS_TIMESTAMP_TYPE() for the type check) before dereferencing.

Copilot · 2026-04-21T09:13:04Z

  SNode* pOrder = ((SOrderByExprNode*)nodesListGetNode(pSelect->pOrderByList, 0))->pExpr;
  if ((QUERY_NODE_TEMP_TABLE == nodeType(pSelect->pFromTable) && isPrimaryKeyImpl(pOrder)) ||
-      (QUERY_NODE_TEMP_TABLE != nodeType(pSelect->pFromTable) && isPrimaryKeyImpl(pOrder))) {
+      (QUERY_NODE_TEMP_TABLE != nodeType(pSelect->pFromTable) && isPrimaryKeyImpl(pOrder)) ||
+      TSDB_DATA_TYPE_TIMESTAMP == ((SExprNode*)pOrder)->resType.type) {
    pSelect->timeLineResMode = TIME_LINE_GLOBAL;


resetResultTimeline() sets TIME_LINE_GLOBAL when the first ORDER BY expression has resType TIMESTAMP, even if the expression is a literal/constant or otherwise does not reference any input column. This can incorrectly treat queries like ORDER BY cast(0 as timestamp) as having a valid timeline, potentially allowing timeline/window functions to run with an undefined row order. Consider requiring the ORDER BY expr to reference at least one column (e.g., nodesExprHasColumn(pOrder)) and/or restricting this to timestamp columns (or expressions derived from columns) rather than constants.

Copilot · 2026-04-21T09:13:04Z

    SNode* pOrder = ((SOrderByExprNode*)nodesListGetNode(pSetOperator->pOrderByList, 0))->pExpr;
-    if (isPrimaryKeyImpl(pOrder)) {
+    if (isPrimaryKeyImpl(pOrder) || TSDB_DATA_TYPE_TIMESTAMP == ((SExprNode*)pOrder)->resType.type) {
      pSetOperator->timeLineResMode = TIME_LINE_GLOBAL;


translateSetOperOrderBy() uses the same resType.type == TIMESTAMP shortcut as resetResultTimeline(), which will also mark UNION/SET queries as TIME_LINE_GLOBAL even when ORDER BY is a constant timestamp expression (no column reference). That can make timeline-dependent functions/windows accept inputs with no well-defined ordering. Apply the same safeguard here (e.g., require nodesExprHasColumn(pOrder) or ensure ORDER BY is a timestamp column/expression derived from columns).

Copilot · 2026-04-21T09:13:04Z

+    // Track first TIMESTAMP column for fallback
+    if (pFirstTsExpr == NULL && TSDB_DATA_TYPE_TIMESTAMP == pExpr->resType.type) {
+      pFirstTsExpr = pExpr;
+    }
+  }
+  // Fallback: when _rowts not found but internal primary key col requested, use first TIMESTAMP col
+  if (!*pFound && isInternalPrimaryKey(pCol) && pFirstTsExpr != NULL) {
+    pFoundExpr = pFirstTsExpr;
+    *pFound = true;


In findAndSetTempTableColumn(), the fallback tracks the first projected expression whose resType is TIMESTAMP, but the requirement/doc says “第一个 TIMESTAMP 列”. As written, a computed/constant TIMESTAMP expression (e.g., now()/cast(...)) could be selected as the fallback timeline, which may not provide a meaningful or stable row order. Consider limiting the fallback to QUERY_NODE_COLUMN timestamp columns (or at least prefer real columns over non-column expressions).

Copilot · 2026-04-21T09:13:04Z

  int32_t code = TSDB_CODE_SUCCESS;
-  if (nodeType(pNode) != QUERY_NODE_COLUMN || !IS_TIMESTAMP_TYPE(getSDataTypeFromNode(pNode)->type) ||
-      !((SColumnNode*)pNode)->isPrimTs) {
+  if (nodeType(pNode) != QUERY_NODE_COLUMN || !IS_TIMESTAMP_TYPE(getSDataTypeFromNode(pNode)->type)) {


checkPrimTS() no longer enforces ((SColumnNode*)pNode)->isPrimTs, which broadens validation for all params marked FUNC_PARAM_MUST_BE_PRIMTS. In this file that attribute is used by elapsed() (intended) but also by the internal _irate_partial signature (see around builtins.c:3033-3046). If _irate_partial still requires an actual primary timestamp, this change will silently relax its contract. Consider introducing a separate param attribute/checker for “any TIMESTAMP column” (used by elapsed) instead of changing the semantics of MUST_BE_PRIMTS globally.

Suggested change

if (nodeType(pNode) != QUERY_NODE_COLUMN || !IS_TIMESTAMP_TYPE(getSDataTypeFromNode(pNode)->type)) {

if (nodeType(pNode) != QUERY_NODE_COLUMN || !IS_TIMESTAMP_TYPE(getSDataTypeFromNode(pNode)->type) ||

!((SColumnNode*)pNode)->isPrimTs) {

feat: 时间线回退功能 - 代码/文档/测试

db54321

parser: 无主键时回退到第一个时间戳列 builtins: 放宽 elapsed 等函数主键检查 docs: 中文需求文档(行为对比/函数分类) test: 54条SQL + 298行期望输出(文件比较模式)

Copilot AI review requested due to automatic review settings April 21, 2026 09:07

facetosea requested review from a team, dapan1121, guanshengliang and zitsen as code owners April 21, 2026 09:07

taosdata-bot Bot added the team engine label Apr 21, 2026

Copilot started reviewing on behalf of facetosea April 21, 2026 09:07 View session

gemini-code-assist Bot reviewed Apr 21, 2026

View reviewed changes

Copilot AI reviewed Apr 21, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: 时间线回退功能 - 代码/文档/测试#35192

feat: 时间线回退功能 - 代码/文档/测试#35192
facetosea wants to merge 1 commit into3.0from
feat/timeline-fallback

facetosea commented Apr 21, 2026

Uh oh!

gemini-code-assist Bot left a comment

Uh oh!

gemini-code-assist Bot Apr 21, 2026

Uh oh!

gemini-code-assist Bot Apr 21, 2026

Uh oh!

gemini-code-assist Bot Apr 21, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Apr 21, 2026

Uh oh!

Copilot AI Apr 21, 2026

Uh oh!

Copilot AI Apr 21, 2026

Uh oh!

Copilot AI Apr 21, 2026

Uh oh!

Copilot AI Apr 21, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants


		## 4. 实现要点

		### 4.1 修改点1：`findAndSetTempTableColumn()` (parTranslater.c)

	if (nodeType(pNode) != QUERY_NODE_COLUMN \|\| !IS_TIMESTAMP_TYPE(getSDataTypeFromNode(pNode)->type)) {
	if (nodeType(pNode) != QUERY_NODE_COLUMN \|\| !IS_TIMESTAMP_TYPE(getSDataTypeFromNode(pNode)->type) \|\|
	!((SColumnNode*)pNode)->isPrimTs) {

Conversation

facetosea commented Apr 21, 2026

Description

Issue(s)

Checklist

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants