kennknowles commented on PR #32165:
URL: https://github.com/apache/beam/pull/32165#issuecomment-2297057635

   I think these are the cases:
   
    - choose v2: we don't even do the v1 translation, and `upload_graph` is 
turned off / does nothing, portable pipeline staged to GCS
    - choose v1: `upload_graph` works and also we will set it automatically if 
necessary
    - leave up to the service: v1 graph follows the usual v1 path and portable 
pipeline staged to GCS
   
   So there might be a corner case for a very large graph that does not choose 
v1 or v2. In this case, we will automatically turn on `upload_graph` and stage 
both the v1 dataflow graph and the portable pipeline to GCS. I am not certain 
if this works. What I said before would suggest it doesn't but I was probably 
wrong. Basically if the staging locations or service-side code paths have 
collisions then it won't work, otherwise it will.


-- 
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