aIbrahiim commented on issue #33859:
URL: https://github.com/apache/beam/issues/33859#issuecomment-4445776742

   I investigated the failure and i think the root cause appears to be this 
change in this pr  #37631 
https://github.com/apache/beam/pull/37631/changes#diff-eb787a9abab531331fb0cd79a99c06746dbb7468fae3b42cd6fe70ea47898894R144
 
   as the failing tests PerKeyOrderingTest.testMultipleStatefulOrdering happens 
in the Python job service while reconstructing coders from the pipeline proto: 
python KeyError: beam:coder:schema:v1
   
   so after this change java started emitting the portable URN:
   
   beam:coder:schema:v1
   
   for SchemaCoder via ModelCoderRegistrar.
   
   however this URN is currently only implemented on the java side but Python 
sdk doesn't register or recognize this coder URN yet so when the ULR (which 
uses the Python job service) receives the pipeline proto, coder deserialization 
fails during lookup
   
   so the regression is effectively:
   
   java emits schema:v1
   python cannot deserialize schema:v1
   
   cc: @acrites  @damccorm @Abacn @Amar3tto 


-- 
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