[lucasfang]

Purpose

Backport three main-branch commits to release-0.4:

• 520acdf79f2e3bef70cd2c9c8512b2119f202822 — correct license attribution and simplify Vortex comments (docs: correct license attribution and simplify Vortex comments #518).
• c84520bb50dca0d5a7de6e083962a88f9e3bd7d9 — align guides with the implementation and document local CRoaring modifications (docs: align guides and CRoaring modification notices #521).
• b1d588761512956f6d4f7e5028d853a2a657d0a0 — validate byte ranges against known file sizes, skip LM payload range queries when caching is disabled, and use actual file lengths in integration-test splits (fix(read): validate byte ranges and honor disabled prefetch cache #522).

The commits retain their original authors and include their source SHAs via cherry-pick -x. Resolved one include conflict in the index integration test by retaining both required filesystem headers. Project and documentation versions remain 0.4.0.

Tests

• pre-commit run --files passed for all 21 changed files.
• git diff --check fdad861537a5e70dd372c184ee1f02f375f89c6c HEAD passed.
• Documentation HTML build passed with warnings treated as errors:

PAIMON_DOCS_VERSION=stable sphinx-build -b html -W --keep-going docs/source /tmp/paimon-release04-docs-html

The native Debug build and all five complete test executables passed (2,722 passed; two conditional skips):

• Common: 1,855 passed; two existing cases skip when caching is disabled.
• Parquet: 240 passed.
• ORC: 120 passed.
• Read integration: 303 passed.
• Write/read with index integration: 204 passed.

cmake --build build-rest --target paimon-common-test paimon-parquet-format-test paimon-orc-format-test paimon-read-inte-test paimon-write-and-read-inte-with-index-test -j16
ctest --test-dir build-rest -R '^paimon-common-test$' --output-on-failure
ctest --test-dir build-rest -R '^paimon-(parquet-format|orc-format)-test$' --output-on-failure -j2
ctest --test-dir build-rest -R '^paimon-(read-inte|write-and-read-inte-with-index)-test$' --output-on-failure -j2

Local validation used ORC/Parquet/Avro/REST with optional Rust formats and sanitizers disabled. The repository-wide unittest target was not run locally; GitHub CI is running the wider build matrix.

API and Format

No changes to installed public API, storage format or protocol. Invalid byte ranges beyond a known file size now fail before splitting or allocating cache entries. File size zero retains its existing unknown-size meaning.

Documentation

Backports corrected guides, license attribution and CRoaring modification notices. No new feature or configuration option.

Generative AI tooling

Generated-by: OpenAI Codex (codex-cli 0.166.0)

[lxy-9602]

+1