PDGGK commented on PR #10529:
URL: https://github.com/apache/seatunnel/pull/10529#issuecomment-4830219677

   Thanks for the careful review, @DanielCarter-stack. Quick responses:
   
   1. `close()` ordering — agreed that wrapping in try-finally so 
`socket.close()` still runs even if `reader.close()` throws is strictly safer; 
happy to make that change. (In practice `reader.close()` closes the underlying 
socket input stream and the following `socket.close()` is idempotent, so the 
socket isn't leaked today, but the try-finally is cleaner.)
   2. `pollNext()` vs `internalPollNext()` — I kept the override minimal to 
stay close to the original, but you're right that implementing 
`internalPollNext()` matches `AbstractSingleSplitReader`'s design (and picks up 
the checkpoint-lock synchronization for free). I can switch to that if 
preferred.
   3. Tests — agreed this is the most valuable follow-up; I can add a 
regression test that opens the reader, calls `pollNext()` twice over a local 
`ServerSocket`, and asserts the second call still reads instead of failing with 
"Socket is closed".
   
   The remaining gate looks like the `Build` check sitting at ACTION_REQUIRED — 
could a committer trigger the full CI run and take a look when convenient? 
Happy to push the above improvements.
   


-- 
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