[Before forward-ports] Antalya 26.3: patch for SettingsChangesHistory

[zvonand]

When porting to future Antalya versions:

• generate .tsv file containing settings values in latest Antalya release (more details in comments in tests/queries/0_stateless/02995_new_settings_history.sh), change the version number in test accordingly;
• add new version section for the next release in src/Core/SettingsChangesHistory.cpp;

When cherry-pinking, do cherry-pick -n avoid picking:

• ...tsv file: it needs to be re-generated
• SettingsChangesHistory.cpp: instead, take Antalya-related sections from previous Antalya release's SettingsChangesHistory.cpp and manually add them to current file. comment all lines with Antalya-specific settings.

Changelog category (leave one):

• Not for changelog (changelog entry is not required)

CI/CD Options

Exclude tests:

• Fast test
• Integration Tests
• Stateless tests
• Stateful tests
• Performance tests
• All with ASAN
• All with TSAN
• All with MSAN
• All with UBSAN
• All with Coverage
• All with Aarch64
• All Regression
• Disable CI Cache

Regression jobs to run:

• Fast suites (mostly <1h)
• Aggregate Functions (2h)
• Alter (1.5h)
• Benchmark (30m)
• ClickHouse Keeper (1h)
• Iceberg (2h)
• LDAP (1h)
• Parquet (1.5h)
• RBAC (1.5h)
• SSL Server (1h)
• S3 (2h)
• S3 Export (2h)
• Swarms (30m)
• Tiered Storage (2h)

[github-actions]

Workflow [PR], commit [9633607]

[ianton-ru]

test_backward_compatibility/test_convert_ordinary.py::test_convert_ordinary_to_atomic broken, fix in #1646 - 24b139f

[ianton-ru]

Regression tests failed because test unmerged features, ignore it.

[ianton-ru]

Does this line correct before feature with object_storage_cluster merged? (Same for all Antalya-only features)

[zvonand]

I guess so. A setting in CH is compared against this file, not the other way around. So it is fine to have some unincluded extra.

[CarlosFelipeOR]

QA Verification — ✅ Approved

PR #1653 (Antalya 26.3 SettingsChangesHistory patch) is safe and did not introduce any regression. This is a forward-port of PR #1465 (merged into antalya-26.1 on 2026-03-02, ~7 weeks of production exposure) — the C++ logic for ClickHouseVersion suffix parsing is byte-for-byte identical to #1465. The only new content in this PR is additional commented-out settings placeholders in SettingsChangesHistory.cpp (no runtime behavior change) and a new 1527-row baseline TSV file for the Antalya canary test. All failing tests in the MasterCI merge run 24640946938 and the PR-CI run are either pre-existing on antalya-26.3, not forward-ported features, or known flaky — none are caused by this PR.

Canary test ✅

02995_new_settings_history (the stateless test directly modified by this PR) passed on 7 build types (amd_asan, amd_coverage, amd_debug, amd_debug s3 storage, amd_tsan, amd_tsan s3 storage, amd_ubsan) and was correctly skipped on arm_binary. No fails.

Summary

Category	Tests / Jobs	Status
Same pre-existing regression failures as PR #1651's MasterCI (identical counts on settings 126 fails, version 4 fails, etc.)	disk_level_encryption, version, settings, session_timezone, aggregate_functions_{2,3}, tiered_storage_*, s3_*_1	✅ Not caused by PR
Features not yet forward-ported to antalya-26.3	swarms, parquet, iceberg_{1,2}, s3_export_partition	✅ Expected
Pre-existing 26.3+ failures (also fail on upstream 26.4)	rbac_1 (check supported timezones), ldap_external_user_directory (empty/missing server)	✅ Not caused by PR
One-off ExpectTimeoutError (infra)	rbac_3 / check alter user add auth methods with two auth methods allowed	✅ Not caused by PR
Known flaky stateless/integration (each affecting 63–144 unrelated PRs over the last 30–60d)	00157_cache_dictionary, test_convert_ordinary_to_atomic, 02415_all_new_functions_must_be_documented, 03443_shared_storage_snapshots, test_cpu_time_fairness[fixed_longer_prd], 03590_simdjson_fallback, 01103_check_cpu_instructions_at_startup, 03706_statistics_preserve_checksums_on_mutations, test_parallel_quorum_actually_parallel, 04043_system_asynchronous_inserts_user_filter, test_kafka_formats_with_broken_message[generate_old_create_table_query], 03634_autopr_output_bytes_estimation, 03927_autopr_input_bytes_estimation_prewhere_filter	✅ Not caused by PR
Caused by this PR	0	🟢

Key evidence

Identical regression-suite pattern as #1651 — comparing the MasterCI runs of #1651 (25120257717) and #1653 (24640946938) on antalya-26.3, the settings job has exactly 126 fails / ~2950 OKs in both, and version has exactly 4 fails / 0 OKs in both. The same pre-existing 26.3 issue, not introduced by either PR.

Forward-port precedent — PR #1465 introduced the same ClickHouseVersion suffix-handling logic in antalya-26.1 on 2026-03-02 with no reported issues since. The C++ diff in this PR is identical; new content is structural-only (commented-out placeholders) with no runtime effect.

[CarlosFelipeOR]

AI audit note: This review comment was generated by AI (gpt-5.3-codex).

Audit update for PR #1653 (SettingsChangesHistory Antalya patch + version parsing update):

Confirmed defects:

• Medium: Malformed compatibility versions are now accepted instead of rejected
  Impact: Typos in compatibility values can silently apply an unintended compatibility profile instead of failing fast, causing hard-to-diagnose settings behavior.
  Anchor: src/Common/ClickHouseVersion.cpp / ClickHouseVersion::ClickHouseVersion
  Trigger: A non-numeric middle component such as 26..1, 26.1., or 26.x.1 in a parsed version string.
  Why defect: The constructor now treats the first non-numeric token (including empty tokens) as suffix and stops parsing, while previous behavior rejected such strings with BAD_ARGUMENTS.
  Fix direction (short): Restrict suffix handling to an explicit terminal suffix pattern; keep empty/intermediate non-numeric components invalid.
  Regression test direction (short): Add parser tests asserting exceptions for malformed inputs and success for valid suffixed forms like 26.1.3.20001.altinityantalya.

• Medium: New stateless check can miss required current-version history updates (false negatives)
  Impact: A setting changed in current Antalya can bypass detection if its name exists in any historical Antalya entry, allowing missing SettingsChangesHistory updates to pass CI.
  Anchor: tests/queries/0_stateless/02995_new_settings_history.sh / filters using position(version, 'altinityantalya') > 0 and Antalya checks without version bound
  Trigger: A setting that was already present in an older Antalya history entry changes again in a newer Antalya release without a new history entry.
  Why defect: The NOT IN exclusion now matches names from all Antalya-tagged versions (or all Session versions in new Antalya blocks), so presence anywhere suppresses detection for current-release deltas.
  Fix direction (short): Bound Antalya checks to the intended version window (e.g., > previous_antalya_version) rather than “any Antalya ever”.
  Regression test direction (short): Add a synthetic test case where a setting changes in two different Antalya versions and ensure the second change is still required.

Coverage summary:

• Scope reviewed: src/Common/ClickHouseVersion.cpp, src/Common/ClickHouseVersion.h, src/Core/SettingsChangesHistory.cpp, tests/queries/0_stateless/02995_new_settings_history.sh, and added baseline tests/queries/0_stateless/02995_settings_26_1_6_20001_antalya.tsv.
• Categories failed: input-validation invariant for version parsing; test oracle strictness for Antalya delta detection.
• Categories passed: call-graph/transition review of compatibility application paths (SettingsImpl::applyCompatibilitySetting, MergeTreeSettings::applyCompatibilitySetting), map ordering/duplicate-key behavior, and static side-effect review of baseline TSV addition.
• Assumptions/limits: static audit only; CI report deep-parser tool could not be executed in this environment because node is unavailable.