Antalya 25.8 backport of #84770: Handle NULLs in ALTER MODIFY COLUMN

[mkmkme]

25.8.15 Stable backport of ClickHouse#84770: Handle NULLs in ALTER MODIFY COLUMN

Same as #1344, but for Antalya 25.8

Changelog category (leave one):

• Backward Incompatible Change

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

ALTER MODIFY COLUMN now requires explicit DEFAULT when converting nullable columns to non-nullable types. Previously such ALTERs could get stuck with cannot convert null to not null errors, now NULLs are replaced with column's default expression (ClickHouse#84770 by @vdimir)

Documentation entry for user-facing changes

...

CI/CD Options

Exclude tests:

• Fast test
• Integration Tests
• Stateless tests
• Stateful tests
• Performance tests
• Aarch64 tests
• All with ASAN
• All with TSAN
• All with MSAN
• All with UBSAN
• All with Coverage
• 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)
• OAuth (5m)
• Parquet (1.5h)
• RBAC (1.5h)
• SSL Server (1h)
• S3 (2h)
• S3 Export (2h)
• Swarms (30m)
• Tiered Storage (2h)

[mkmkme]

The tests passed locally. Let's see if something breaks in CI. After that I'll self-verify

[github-actions]

Workflow [PR], commit [9c9555d]

[mkmkme]

@blau-ai

[blau-ai]

CI triage — @mkmkme

Verdict: 2 distinct failures, both NOT caused by this PR (1 pre-existing regression-suite failure, 1 CI-infra failure). All build/stateless/integration/stress checks that actually exercise this PR's code are green. Nothing to fix in the PR itself.

This PR touches only the nullable→non-nullable conversion path for ALTER MODIFY COLUMN (inplaceBlockConversions.cpp, AlterCommands.cpp, IMergeTreeReader.cpp). None of the failing checks exercise that code.

---

1. RegressionTestsRelease / Iceberg (1) & Iceberg (2) (+ the Regression x86 iceberg_1/2 summary check-runs) — pre-existing, not PR-related

Decisive log excerpts:

✘ [ Fail ]  /iceberg/iceberg engine/rest catalog/position delete reads/read position deletes  (AssertionError)
✘ [ Error ] /iceberg/iceberg engine/rest catalog/sort key timezone   (AttributeError: feature_supported)
✘ [ Fail ]  /iceberg/iceberg engine/glue catalog/position delete reads/read position deletes  (AssertionError)
✘ [ Fail ]  /iceberg/iceberg table engine/feature/position delete smoke   (AssertionError)
✘ [ Fail ]  /iceberg/iceberg table engine/feature/compaction smoke        (AssertionError)

Evidence it's pre-existing / unrelated:

• The failures are all in the Iceberg suite (position-delete reads, position-delete smoke, compaction smoke, sort-key-timezone). Iceberg storage does not go through IMergeTreeReader / ALTER MODIFY COLUMN, so this PR cannot affect them.
• The same failures reproduce identically on other open backport PRs against the same base (antalya-25.8) that touch completely different code:
  • Antalya 25.8 backport of #90243: Fix crash in StorageDistributed when parsing malformed shard directory names #2168 (StorageDistributed shard-dir parsing) — Iceberg (1) errored, Iceberg (2) failed, same 1 module (1 errored) / 1 module (1 failed) signature.
  • Antalya 25.8 Backport of #90059: Unblock ttl part drops for cold volumes #2172 (TTL part drops for cold volumes) — same signature.
• The sort key timezone case is an AttributeError: feature_supported — a bug in the clickhouse-regression test framework itself, not in the ClickHouse binary.

Next step: Safe to disregard for merging this PR. These belong to the Iceberg regression suite's known/flaky set on antalya-25.8; worth a separate ticket to the regression-suite owners if not already tracked, but they are not a blocker introduced here.

---

2. FinishCIReport — CI infra, not PR-related (already being fixed separately)

Decisive log excerpt:

WARNING:Failed to fetch workflow config from
 https://.../PRs/2169/<sha>/config_workflow/workflow_config_pr.json: 404
##[error]Process completed with exit code 1.

The report generator (create_workflow_report.py) aborts because it 404s on workflow_config_pr.json (the Config Workflow job was skipped, so the artifact was never uploaded). This is a report-tooling/infra issue, unrelated to the code change.

Next step: This exact failure is already being addressed in #2174 "Antalya 25.8 - fix config url for report job". Once #2174 merges, rebasing/re-running this PR should clear FinishCIReport. No PR change needed here.

---

Health check

The PR's substantive checks are green: all Builds (amd/arm, release/debug/binary), Fast test, Stateless tests (debug parallel/sequential, s3/DatabaseReplicated/ParallelReplicas variants), Integration tests (amd_binary 1–5/5), and Stress test (amd_debug) all passed. The two red checks are an unrelated pre-existing Iceberg suite failure and an infra report-generation bug fixed elsewhere — so from a correctness standpoint this backport looks clean and mergeable once the infra check is unblocked.

No code changes proposed — nothing here is a PR-caused failure.