You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flink.apache.org by "featzhang (Jira)" <ji...@apache.org> on 2026/02/14 06:27:00 UTC

[jira] [Created] (FLINK-39090) [SQL] Enhanced join condition structured display

featzhang created FLINK-39090:
---------------------------------

             Summary: [SQL] Enhanced join condition structured display
                 Key: FLINK-39090
                 URL: https://issues.apache.org/jira/browse/FLINK-39090
             Project: Flink
          Issue Type: Improvement
          Components: Table SQL / Planner
            Reporter: featzhang


Currently, the EXPLAIN output for FlinkLogicalJoin shows join conditions using positional references (e.g., =($0, $3)), which makes it difficult for users to understand the actual join logic, especially in complex queries with multiple tables.

Example of current output:
{code:java}
Join(condition=[=($0, $3)])
{code}
This is not user-friendly because:
1. Users need to manually map positional indexes to field names
2. In complex queries with many fields, it's error-prone
3. It reduces SQL debuggability and readability

Proposed enhancement:
Display join conditions with actual field names in a structured format:
{code:java}
Join:
  Type: INNER
  Condition: orders.user_id = users.id
{code}
Benefits:
 - Improved user experience and query debugging
 - Better alignment with SQL semantics
 - Easier to understand join logic at a glance
 - No impact on RelNode structure or runtime performance

Implementation approach:
 - Override explainTerms method in FlinkLogicalJoin class
 - Parse RexNode conditions recursively
 - Resolve field references to actual field names
 - Support EQUALS, AND, OR operators
 - Fallback to default formatting for complex expressions

This is a display-only enhancement and does not modify any query optimization logic.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)