SEPURI-SAI-KRISHNA commented on PR #13092:
URL: https://github.com/apache/seatunnel/pull/13092#issuecomment-6844708560

   @nzw921rx Thanks, done in `65703fe498`, following the CDC split column 
selection:
   
   1. When the split key is chosen automatically, a nullable primary key or 
unique key column is skipped with a WARN, and the table is read as one split if 
no column is left.
   2. An explicit `partition_column` is used as configured. Its option 
description and the split key section of the Jdbc source docs (EN and ZH) now 
say that rows with NULL in it are not read.
   
   The NULL split, its probe query and the `JdbcSourceSplit` change are 
removed, so the PR is down to 7 files. `JdbcMysqlSplitIT` has a MySQL case for 
point 1 (one split with only a nullable unique key, the NOT NULL unique key 
when there is one); with the check removed it reads 90 of 100 rows. The PR 
description and the `incompatible-changes.md` entry are updated. Could you take 
another look when you have time?
   
   @SEZ9 Thanks for the review. With this change the remaining items no longer 
apply: there is no new split to restore with an older jar, no `IS NOT NULL` on 
hash splits, no `isNullableSplitKey`, and `JdbcSourceSplit` is unchanged. My 
earlier comment is complete on GitHub; it ended by asking @nzw921rx to confirm 
this scope, which the comment above does.
   
   @DanielLeens this replaces the version you approved, so a new look would 
help too.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to