steveloughran commented on pull request #2735:
URL: https://github.com/apache/hadoop/pull/2735#issuecomment-789011590


   Quick review of this, especially the factoring out
   
   Good: 
   * unified logic in one place
   Bad
   * it's still complex
   * it's not easy for S3A to optimise further
   
   Looking at s3a rename I now want
   
   * getFileStatus on source file to return the metadata so that there's no 
second call on the copy
   * probe for parent dir to know it expects a dir. e.g callback 
`getDirStatus(dir)`. 
   * if the source is a dir, to initiate a LIST and have that feed straight 
into the object scan. That's tricky, but you know...
   * option to turn off the check for the dir existing.
   * async fetch of all the probes
   
   so: all the callbacks would be completable futures; default impl would be 
sequential, but for the stores we can go async.
   
   + split into API (o.a.h.fs.api) and impl and have something in API To be the 
factory for the impl: downstream classes do not need to look into .impl.
   
   No timetable for this.
   
   


----------------------------------------------------------------
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:
[email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to