aljoscha commented on pull request #12132:
URL: https://github.com/apache/flink/pull/12132#issuecomment-629851215


   That was not very nice. I assumed it was clear that it is important that the 
upgrade/migration test be added separately before the changes to the 
serializer. As it was merged on master this is not reproducible, because there 
is no code for generating the v1 snapshots and both the changes to the 
test/snapshots/serializer itself happen in one commit. Also, the special 
`RecoverableWriter` in `TestUtils` meant that we use special code paths for a 
test, which was not needed in my version. And relying on custom file 
copying/cleanup in tests makes them more susceptible to subtle problems or 
leftover junk.
   
   Plus, there are still some issues in the code, such as raw-use of generic 
types. For example in most places where a `BucketStateSerializer` is created. 
And this is a real issue because the interfaces of what `BucketStateSerializer` 
expects and what `WriterProperties.getInProgressFileRecoverableSerializer()` 
returns do not match, even though they should.
   
   I reverted the changes on `master` and added the `BucketStateSerializer` as 
a commit. I also fixed the problem that made the test fail for the v2 
serializer and created an updated branch that has all your changes on top of 
the reverts, with the code fixed to pass the test: 
https://github.com/aljoscha/flink/commits/pr-12132-file-sink. This could be 
merged, but as I said there are still some other issues in the code.
   
   I understand that we're under time pressure but rushed solo efforts are not 
the way to go here, especially if two reviewers still had comments/outstanding 
suggestions. This now caused me a lot of extra work and you have to go and look 
at things again and merge them.
   
   
   


----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


Reply via email to