Repository navigation
[core] Reject hidden-leading partition values in value-only layout - #10240
Closed
LuciferYang wants to merge 2 commits into
Closed
LuciferYang wants to merge 2 commits into
LuciferYang wants to merge 2 commits into
Conversation
In a value-only partition layout the write validator rejected only null, empty, '.' and '..' values, and the escaping kept '_' literal, so a value like '_x' or '.x' was written to a directory whose name starts with '_' or '.'. The read path treats such directories as hidden staging or metadata directories and skips them, so the partition and all its data silently disappeared from scans. Reject values whose name would be hidden, except the table's configured default partition name, which scans un-hide explicitly. The configured name is threaded through the format-table writers, commit, partition-path resolver and registry validator, the spark partition repair and the HiveCatalog partition locations; callers without a table context fall back to the well-known default name. Assisted-by: GLM-5.3
…mat-table partition DDL
LuciferYang
marked this pull request as draft
October 2, 2026 03:19
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Purpose
In a value-only partition layout the write validator rejected only null, empty, "." and "..", so a value such as "x" or ".tmp" was written to a directory whose name starts with "" or ".". The read path treats those directories as hidden and skips them, so the partition and all of its data silently disappeared from scans.
This rejects a partition value whose rendered value-only path segment would be hidden, so the write fails fast instead of producing an unreadable partition. The only exempt value is the table's configured default partition name, because scans un-hide exactly that name (the built-in default
__DEFAULT_PARTITION__itself starts with "_"). The configured name is threaded through the format-table writers, commit, partition-path resolver, registry validator, stats collector, scan, the HiveCatalog partition locations, and the Spark partition DDL paths (ADD / DROP / TRUNCATE / ANALYZE PARTITION and ADD PARTITION ... LOCATION), so a table whosepartition.default-nameis itself a hidden name can still write and manage its null partition rather than having that DDL rejected. Callers without a table context fall back to the well-known default name. Key-value layout is unchanged, since the component there is "pt=_x" and does not start with a hidden prefix.Tests
New cases in
PartitionPathUtilsTestpin: value-only "_x" and ".x" are rejected; the built-in default partition name and a configured custom default name stay writable while other hidden-leading values remain rejected; and a key-value "pt=_x" component stays visible.API and Format
No.
Documentation
No.