You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hudi.apache.org by "Davis Zhang (Jira)" <ji...@apache.org> on 2025/07/13 19:21:00 UTC
[jira] [Created] (HUDI-9587) Secondary index design recap - Is it a good choice to track everything as String?
Davis Zhang created HUDI-9587:
---------------------------------
Summary: Secondary index design recap - Is it a good choice to track everything as String?
Key: HUDI-9587
URL: https://issues.apache.org/jira/browse/HUDI-9587
Project: Apache Hudi
Issue Type: Bug
Components: index
Reporter: Davis Zhang
Fix For: 1.2.0
context https://issues.apache.org/jira/browse/HUDI-9566
The issue is record in data table can be of any data type
The record value tracked by secondary index is all converted to string via "toString" call.
What make things worse is the toString call can be invoked on various data type, based on how the read path and write path prepare keys to look up / updated keys for SI update.
We need to recap the design. For SI index to work, the acceptance criteria is
record matching mechanism in data table should be the same as record matching mechanism in the MDT.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)