DanielLeens commented on issue #12929:
URL: https://github.com/apache/seatunnel/issues/12929#issuecomment-6315602391

   Thanks for the precise reproduction and the historical UID analysis.
   
   Classification: F / Transform-V2 serialization compatibility. There is now 
one implementation lane: #12938. It explicitly pins the pre-#12804 
`AbstractMultiCatalogTransform` UID used by the archived 
`legacy-boolean-wrapper` fixture and makes the construction-only 
`transformLevelConfig` transient.
   
   The scope is reasonable for restoring the archived wrapper test, but the PR 
description also correctly notes an important compatibility boundary: streams 
written by the interim post-#12804 class with its implicit UID are not 
compatible with the proposed historical UID. Please keep the compatibility 
decision and implementation discussion on #12938; do not open a competing 
source change here.
   
   Before this issue can close, #12938 needs to demonstrate both contracts 
explicitly:
   
   1. the archived pre-#12804 wrapper deserializes and still exercises the 
expected LLM boolean behavior;
   2. a wrapper serialized by the patched current code round-trips correctly, 
including the existing validation-mode coverage; and
   3. a regression asserts the superclass UID so a future structural change 
cannot silently break the archived contract again.
   
   Please also state in the PR whether serialized transform wrappers are 
persisted across application upgrades. That determines whether the 
interim-stream incompatibility is only a development-fixture boundary or needs 
migration/release treatment. The issue will remain open until the PR is 
reviewed and merged.


-- 
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