Support SQL standard VALUES clause as a table expression by Desel72 · Pull Request #100143 · ClickHouse/ClickHouse

Desel72 · 2026-03-19T22:08:02Z

Adds support for the SQL standard VALUES clause syntax: SELECT * FROM (VALUES (1, 'a'), (2, 'b')) AS t(id, val)

This is rewritten internally to use the existing values table function, so all existing type inference and column naming logic is reused.

Closes #99605

Changelog category (leave one):

New Feature

Changelog entry (a user-readable short description of the changes that goes into CHANGELOG.md):

Support SQL standard VALUES clause as a table expression in FROM, e.g. SELECT * FROM (VALUES (1, 'a'), (2, 'b')) AS t(id, val).

Documentation entry for user-facing changes

Documentation is written (mandatory for new features)

Adds support for the SQL standard `VALUES` clause syntax: `SELECT * FROM (VALUES (1, 'a'), (2, 'b')) AS t(id, val)` This is rewritten internally to use the existing `values` table function, so all existing type inference and column naming logic is reused. Closes ClickHouse#99605 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

clickhouse-gh · 2026-03-19T22:14:46Z

Workflow [PR], commit [954d1d3]

Summary: ❌

job_name	test_name	status	info	comment
BuzzHouse (amd_msan)		failure
	MemorySanitizer: use-of-uninitialized-value (STID: 3229-5315)	FAIL	cidb, issue	ISSUE CREATED

AI Review

Summary

This PR adds support for SQL-standard VALUES in FROM by rewriting (VALUES ...) into an internal SQLStandardValues table function. The implementation direction is good, but there are blockers around rollout policy and one parser-semantic issue: the feature is enabled unconditionally (no experimental gate), and the parser currently accepts non-standard row forms like (VALUES 1, 2).

PR Metadata

⚠️ Changelog category is currently New Feature, but this change is intended as experimental (docs mention allow_experimental_sql_standard_values_clause). It should be Experimental Feature.
⚠️ Changelog entry text is clear, but it should explicitly state experimental status and how to enable it.

Exact replacement:

Changelog category: - Experimental Feature
Changelog entry: Add experimental support for SQL-standard \ \VALUES` in
`FROM` (e.g.
`SELECT * FROM (VALUES (1, 'a')) AS t(id, val)`). Enable with
`allow_experimental_sql_standard_values_clause = 1`.`

Missing context

⚠️ No CI report/log links were provided in the review input, so runtime/regression signals could not be cross-checked.

Findings

❌ Blockers
- [src/Parsers/ExpressionElementParsers.cpp:194] SQL-standard VALUES support is introduced unconditionally in parser logic, but this is a new behavior and should be guarded behind an experimental setting for safe rollout.
  - Suggested fix: gate this branch by allow_experimental_sql_standard_values_clause (or equivalent parser-accessible setting) and keep default behavior unchanged when disabled.
⚠️ Majors
- [src/Parsers/ExpressionElementParsers.cpp:202] The parser uses ParserExpression directly for each VALUES item, so non-standard forms like (VALUES 1, 2) are accepted. SQL-standard form requires row constructors.
  - Suggested fix: require each item to be a parenthesized row expression before adding it to SQLStandardValues arguments.
- [docs/en/sql-reference/statements/select/from.md:24] The new example shows VALUES usage without clearly stating/enforcing the experimental gate, which can mislead users.
  - Suggested fix: add explicit note or setup line with SET allow_experimental_sql_standard_values_clause = 1;.
- [docs/en/sql-reference/table-functions/values.md:198] Docs still say rewrite goes to values, while code rewrites to SQLStandardValues.
  - Suggested fix: update wording to reflect the actual internal rewrite target.

ClickHouse Rules

Item	Status	Notes
Deletion logging	➖
Serialization versioning	➖
Core-area scrutiny	✅
No test removal	✅
Experimental gate	❌	New parser behavior is enabled by default
No magic constants	✅
Backward compatibility	⚠️	New syntax behavior changes parser acceptance unconditionally
`SettingsChangesHistory.cpp`	➖
PR metadata quality	⚠️	Category/entry should reflect experimental rollout
Safe rollout	⚠️	Missing feature flag gate in implementation
Compilation time	✅

Final Verdict

Status: ⚠️ Request changes
Minimum required actions:
- Add an experimental gate for SQL-standard VALUES parser behavior.
- Tighten parser to require row-constructor form for VALUES items (or explicitly document and justify broader grammar).
- Align docs and PR metadata with actual rollout behavior.

alexey-milovidov

The code looks good.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…ix/issue-99704-v2

Desel72 · 2026-03-20T17:54:20Z

Hi @alexey-milovidov
I think failure isn't related with me. Could you please review this?

alexey-milovidov · 2026-03-20T18:41:38Z

@Desel72, it shows that some additions to the docs are needed for a new feature. At least mention this feature in the docs in the appropriate place.

…ix/issue-99704-v2

…House into fix/issue-99704-v2

…ix/issue-99704-v2 # Conflicts: # ci/jobs/scripts/check_style/aspell-ignore/en/aspell-dict.txt

…House into fix/issue-99704-v2 # Conflicts: # ci/jobs/scripts/check_style/aspell-ignore/en/aspell-dict.txt

…rd_values_clause` setting Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…alSettings, fix aspell dict merge conflict

clickhouse-gh · 2026-03-22T11:37:06Z

+-- https://github.com/ClickHouse/ClickHouse/issues/99605
+
+SET enable_analyzer = 1;
+SET allow_experimental_sql_standard_values_clause = 1;


Please add targeted negative/regression coverage for the parser rewrite edge cases:

A query that validates a first-row single-string literal is treated as row data (not as values structure), e.g. (VALUES ('x UInt8'), (2)).

A query that asserts the syntax is rejected when allow_experimental_sql_standard_values_clause = 0.

Right now the test only covers happy-path enabled usage, so these two important failure modes can regress silently.

…ix/issue-99704-v2

…House into fix/issue-99704-v2

clickhouse-gh · 2026-03-24T00:54:47Z

 DECLARE(Bool, allow_experimental_ytsaurus_dictionary_source, false, R"(
    Experimental dictionary source for integration with YTsaurus.
    )", EXPERIMENTAL) \
+    DECLARE(Bool, allow_experimental_sql_standard_values_clause, false, R"(


The PR marks this as New Feature, but this syntax is explicitly gated by allow_experimental_sql_standard_values_clause. Please switch the PR template Changelog category to Experimental Feature to match ClickHouse rollout policy.

Avogar · 2026-03-24T12:36:17Z

 DECLARE(Bool, allow_experimental_ytsaurus_dictionary_source, false, R"(
    Experimental dictionary source for integration with YTsaurus.
    )", EXPERIMENTAL) \
+    DECLARE(Bool, allow_experimental_sql_standard_values_clause, false, R"(


I don't know why AI decided to ask you to put it under experimental setting. The feature is not that big and complex to do it. Let's remove the setting and allow this syntax out of the box. Ignore AI comments about experimental setting

Please, remove this setting, as I said, it's not needed

…ix/issue-99704-v2

… edge case tests

… and setting gate

…tion check

clickhouse-gh · 2026-03-24T19:27:34Z

    Experimental dictionary source for integration with YTsaurus.
    )", EXPERIMENTAL) \
+    DECLARE(Bool, allow_experimental_sql_standard_values_clause, false, R"(
+Enables support for the SQL standard VALUES clause as a table expression in FROM, e.g. SELECT * FROM (VALUES (1, 'a'), (2, 'b')) AS t(id, val). Internally rewritten to the existing values table function.


The setting description says the syntax is rewritten to the existing values table function, but this PR now rewrites to SQLStandardValues. Please update this text to match the actual implementation to avoid confusing users and future maintainers.

Suggested wording: "Internally rewritten to the internal SQLStandardValues table function."

…lizer

Avogar · 2026-03-25T13:45:04Z

 DECLARE(Bool, allow_experimental_ytsaurus_dictionary_source, false, R"(
    Experimental dictionary source for integration with YTsaurus.
    )", EXPERIMENTAL) \
+    DECLARE(Bool, allow_experimental_sql_standard_values_clause, false, R"(


Please, remove this setting, as I said, it's not needed

Avogar · 2026-03-25T13:53:27Z

+/// SQL standard VALUES clause: (VALUES (1, 'a'), (2, 'b'))
+/// Rewritten by the parser to SQLStandardValues(tuple(1, 'a'), tuple(2, 'b'))
+/// This checks the experimental setting and never interprets the first arg as schema.
+class TableFunctionSQLStandardValues : public ITableFunction


We can avoid creating new class for this, just make TableFunctionValues template with a bool flag:

template <bool interpret_first_argument_as_structure> class TableFunctionValues : public ITableFunction

And check this template argument in all appropriate places, also function name can depend on templat argument.

…put getTypesFromArgument in anonymous namespace

clickhouse-gh · 2026-03-25T18:02:14Z

+        ParserExpression expr_parser;
+
+        ASTPtr value_expr;
+        if (!expr_parser.parse(pos, value_expr, expected))


ParserExpression here accepts any expression, so (VALUES 1, 2) is parsed successfully even though SQL-standard VALUES requires row constructors (expr, ...) per row.

This makes the new syntax less strict than documented and can introduce parsing ambiguities with future grammar extensions. Please require each item after VALUES to be a parenthesized row expression (or at minimum reject non-parenthesized items) before rewriting to SQLStandardValues.

clickhouse-gh · 2026-03-25T20:44:42Z

LLVM Coverage Report

Metric	Baseline	Current	Δ
Lines	84.10%	84.10%	+0.00%
Functions	24.50%	24.60%	+0.10%
Branches	76.70%	76.70%	+0.00%

PR changed lines: PR changed-lines coverage: 97.18% (69/71, 0 noise lines excluded)
Diff coverage report
Uncovered code

Desel72 · 2026-03-25T23:22:07Z

@Avogar Thanks
MSan failure is in FileSegment::getDownloadedSize called from CachedOnDiskWriteBufferFromFile::appendFilesystemCacheLog during CREATE TABLE. This is a pre-existing issue in the filesystem cache layer, unrelated to the parser/table function changes in this PR.

Desel72 · 2026-03-26T00:52:04Z

Thanks @Avogar

zlareb1 · 2026-03-26T11:25:31Z

@Desel72 small doc update required #100791

alexey-milovidov added the can be tested Allows running workflows for external contributors label Mar 19, 2026

clickhouse-gh Bot added the pr-feature Pull request with new product feature label Mar 19, 2026

alexey-milovidov approved these changes Mar 19, 2026

View reviewed changes

alexey-milovidov self-assigned this Mar 19, 2026

clickhouse-gh Bot reviewed Mar 19, 2026

View reviewed changes

Comment thread src/Parsers/ExpressionElementParsers.cpp

Desel72 and others added 3 commits March 20, 2026 00:18

Add documentation for SQL standard VALUES clause

08c3fd7

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Update 04049_values_clause_standard_sql.sql

6828b81

Merge branch 'master' of https://github.com/Desel72/ClickHouse into f…

634c9a5

…ix/issue-99704-v2

Desel72 added 4 commits March 20, 2026 23:10

Merge branch 'master' of https://github.com/Desel72/ClickHouse into f…

78ff88a

…ix/issue-99704-v2

Merge branch 'fix/issue-99704-v2' of https://github.com/Desel72/Click…

e58993e

…House into fix/issue-99704-v2

Add DuckDB to aspell dictionary

50e5601

Merge branch 'master' into fix/issue-99704-v2

920c8f9

clickhouse-gh Bot reviewed Mar 21, 2026

View reviewed changes

Comment thread src/Parsers/ExpressionElementParsers.cpp

Desel72 and others added 4 commits March 21, 2026 23:36

Merge branch 'master' of https://github.com/Desel72/ClickHouse into f…

6780d90

…ix/issue-99704-v2 # Conflicts: # ci/jobs/scripts/check_style/aspell-ignore/en/aspell-dict.txt

Merge branch 'fix/issue-99704-v2' of https://github.com/Desel72/Click…

bb53f91

…House into fix/issue-99704-v2 # Conflicts: # ci/jobs/scripts/check_style/aspell-ignore/en/aspell-dict.txt

Gate SQL standard VALUES clause behind `allow_experimental_sql_standa…

d337e40

…rd_values_clause` setting Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Fix CI: add setting to SettingsChangesHistory and enableAllExperiment…

69395f1

…alSettings, fix aspell dict merge conflict

clickhouse-gh Bot reviewed Mar 22, 2026

View reviewed changes

Comment thread docs/en/sql-reference/statements/select/from.md

alexey-milovidov added 2 commits March 22, 2026 12:27

Merge remote-tracking branch 'origin/master' into update-pr-100143

44d4b35

Merge remote-tracking branch 'origin/master' into update-pr-100143

d826777

clickhouse-gh Bot reviewed Mar 22, 2026

View reviewed changes

Desel72 added 4 commits March 23, 2026 18:09

Merge branch 'master' into fix/issue-99704-v2

4852f30

Merge branch 'master' of https://github.com/Desel72/ClickHouse into f…

430df61

…ix/issue-99704-v2

Add negative/regression test coverage for VALUES clause edge cases

74058db

Merge branch 'fix/issue-99704-v2' of https://github.com/Desel72/Click…

6d462fb

…House into fix/issue-99704-v2

clickhouse-gh Bot reviewed Mar 24, 2026

View reviewed changes

Avogar self-assigned this Mar 24, 2026

Avogar reviewed Mar 24, 2026

View reviewed changes

Desel72 added 3 commits March 24, 2026 13:36

Merge branch 'master' of https://github.com/Desel72/ClickHouse into f…

d778f03

…ix/issue-99704-v2

Fix VALUES clause: skip schema detection for SQL standard syntax, add…

ec83fd1

… edge case tests

Use separate sqlStandardValues table function to fix schema detection…

385ff3c

… and setting gate

clickhouse-gh Bot reviewed Mar 24, 2026

View reviewed changes

Comment thread docs/en/sql-reference/table-functions/values.md

Desel72 added 3 commits March 24, 2026 15:02

Fix style: use SQL in all caps in class and function names

783e106

Remove unnecessary tuple wrapping that breaks single-column VALUES

c1355b7

Add description to SQLStandardValues table function to pass documenta…

00031fa

…tion check

clickhouse-gh Bot reviewed Mar 24, 2026

View reviewed changes

Fix build: add missing category field to FunctionDocumentation initia…

27abdbf

…lizer

Avogar reviewed Mar 25, 2026

View reviewed changes

Refactor: template TableFunctionValues, remove experimental setting, …

954d1d3

…put getTypesFromArgument in anonymous namespace

clickhouse-gh Bot reviewed Mar 25, 2026

View reviewed changes

Avogar added this pull request to the merge queue Mar 25, 2026

Merged via the queue into ClickHouse:master with commit 438e5c9 Mar 25, 2026
151 of 153 checks passed

robot-ch-test-poll2 added the pr-synced-to-cloud The PR is synced to the cloud repo label Mar 26, 2026

zlareb1 mentioned this pull request Mar 26, 2026

Documentation for SQL standard VALUES clause references non-existent setting allow_experimental_sql_standard_values_clause #100791

Closed

Blargian mentioned this pull request Mar 26, 2026

Docs: remove unneeded experimental setting for SQL standard VALUES clause as a table expression #100795

Merged

1 task

Conversation

Desel72 commented Mar 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changelog category (leave one):

Changelog entry (a user-readable short description of the changes that goes into CHANGELOG.md):

Documentation entry for user-facing changes

Uh oh!

clickhouse-gh Bot commented Mar 19, 2026 • edited by Avogar Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

AI Review

Summary

PR Metadata

Missing context

Findings

ClickHouse Rules

Final Verdict

Uh oh!

alexey-milovidov left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Desel72 commented Mar 20, 2026

Uh oh!

alexey-milovidov commented Mar 20, 2026

Uh oh!

Uh oh!

Uh oh!

clickhouse-gh Bot Mar 22, 2026

Choose a reason for hiding this comment

Uh oh!

clickhouse-gh Bot Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

Avogar Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

Avogar Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

clickhouse-gh Bot Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

Avogar Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Avogar Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

clickhouse-gh Bot Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

clickhouse-gh Bot commented Mar 25, 2026

LLVM Coverage Report

Uh oh!

Desel72 commented Mar 25, 2026

Uh oh!

Uh oh!

Desel72 commented Mar 26, 2026

Uh oh!

zlareb1 commented Mar 26, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Desel72 commented Mar 19, 2026 •

edited

Loading

clickhouse-gh Bot commented Mar 19, 2026 •

edited by Avogar

Loading