Antalya 26.1: Fix view over iceberg

[ianton-ru]

Changelog category (leave one):

• Bug Fix (user-visible misbehavior in an official stable release)

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

Fix incorrect work view over Iceberg table

Documentation entry for user-facing changes

Solved #1669 - do not make null values for complex column names like materialized(time).
Solved #1629 - use ParquetReadRowGroups in test instead of S3GetObject/AzureGetObject.

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 [3cd97c6]

[ianton-ru]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 024385ff6f

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Guard expression parsing for non-SQL-safe column names

parseQuery(parser, column_name, ...) throws on syntax errors, but column_name here comes from getNameInStorage() (raw storage names), not guaranteed to be a standalone SQL expression. When schema evolution leaves a nullable column absent in some files and that column name requires quoting (for example names with spaces/symbols), this new path raises SYNTAX_ERROR and aborts the read instead of applying the NULL-constant fallback.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Avoid treating arbitrary parsed expressions as generated columns

Using expr->as<ASTFunction>() as the skip condition is too broad: quoted physical column names like a-b are parsed as expression ASTs, so this branch skips NULL-constant synthesis for genuinely missing nullable columns. In evolved Iceberg schemas, that leaves such columns in requested_columns_copy, and later readers may try to fetch non-existent parquet fields instead of returning NULLs.

Useful? React with 👍 / 👎.

[ianton-ru]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. 👍

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[ianton-ru]

PR #1671 — Antalya 26.1: Fix view over Iceberg

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

Scope: Altinity fork, branch vs antalya-26.1 — StorageObjectStorageSource.cpp, tests/integration/test_storage_iceberg_with_spark/test_read_constant_columns_optimization.py.

---

Confirmed defects

No confirmed defects in reviewed scope.

(Earlier revision had orphaned parser includes after dropping parseQuery; the current branch tip removes those includes, and the net diff vs antalya-26.1 adds only the materialize(…) heuristic in createReader.)

---

Coverage summary

Item	Detail
Scope reviewed	Iceberg experimental read optimization path that synthesizes NULL constants for nullable columns missing from per-file metadata, and skips that for names shaped like materialize(...) (view-over-Iceberg). Integration test changes: ParquetReadRowGroups assertions and new view test.
Categories failed	—
Categories passed	No SQL parse of raw storage names (no throw on odd names); no broad “any function AST” skip; includes unchanged vs base in net diff; changed logic uses only local state (no new shared-state / locking concerns in the diff hunk).

---

Call graph (in-scope)

Entry: StorageObjectStorageSource::createReader
Branch: allow_experimental_iceberg_read_optimization → file metadata present → iterate columns_info (constants / all-null) → second loop over requested_columns_list keyed by getNameInStorage()
Effect: Optional constant_columns / constant_columns_with_values → requested_columns_copy.eraseNames → downstream pipe / count fast path.

---

Residual risks (not confirmed defects)

1. Heuristic breadth: Only names matching materialize( prefix and ) suffix are skipped. If another computed storage name can appear in the same nullable-missing-metadata situation, wrong NULL synthesis remains theoretically possible; no in-repo reproduction was traced in review.
2. Tests: Assertions use ParquetReadRowGroups and assume one row group per data file as in the fixture. check_events accepts is_cluster but the SQL does not use it (harmless, slightly misleading API).

---

Changelog alignment (from PR description)

• Addresses incorrect behavior for views over Iceberg when complex names like materialize(time) must not be forced to NULL-from-missing-field.
• Tests: prefer ParquetReadRowGroups over object-store get-object profile events where appropriate.

---

References

• PR: Antalya 26.1: Fix view over iceberg #1671
• Related issues cited on PR: Antalya 26.1, View over Iceberg table returns NULLs instead of values #1669, Antalya 26.1 - tests test_storage_iceberg_with_spark/test_read_constant_columns_optimization.py are unstable #1629

[ianton-ru]

Failed tests looks unrelated

[ianton-ru]

Rebase on 26.3: #1759