fix: colDataKeepFirstNRows reset varmeta.length when all kept rows ar…

[facetosea]

…e NULL

Description

Issue(s)

• Close/close/Fix/fix/Resolve/resolve: Issue Link

Checklist

Please check the items in the checklist if applicable.

• Is the user manual updated?
• Are the test cases passed and automated?
• Is there no significant decrease in test coverage?

[Copilot]

Pull request overview

This PR fixes a variable-length column trimming edge case in blockDataKeepFirstNRows/colDataKeepFirstNRows where keeping only NULL rows could leave varmeta.length stale, and adds a regression test to cover LIMIT/OFFSET scenarios that exercise that path.

Changes:

• Reset SColumnInfoData::varmeta.length to 0 when the kept range contains no non-NULL var-length values.
• Add a Python system test covering LIMIT and LIMIT+OFFSET on tables where the first N var-length values are all NULL.
• Clarify/adjust numeric_limits<intx::uint<N>>::digits10 computation documentation and constant precision.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
source/common/src/tdatablock.c	Resets varmeta.length to 0 when all kept var-length rows are NULL to avoid stale length usage after LIMIT trimming.
test/cases/01-DataTypes/test_null_column.py	Adds regression coverage for LIMIT/LIMIT+OFFSET on var-length columns with leading NULLs.
source/libs/decimal/src/detail/intx/int128.hpp	Documents digits10 derivation and uses a more precise log10(2) constant in the constexpr calculation.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[gemini-code-assist]

Code Review

This pull request addresses an issue where variable-length columns could retain stale length values when all rows in a data block are NULL following a LIMIT operation. It ensures varmeta.length is reset to 0 in such cases and replaces a fatal error with a debug log. Additionally, the precision of the digits10 constant in the decimal library was improved, and a new regression test was added. Review feedback suggests refining the condition check for NULL offsets and verifying format specifiers in logging to prevent potential sign-mismatch issues.

[dapan1121]

LGTM