tom-s-powell commented on issue #39687:
URL: https://github.com/apache/arrow/issues/39687#issuecomment-1899959042

   So I did actually manage to avoid fully going to `VectorSchemaRoot`, by 
copying some of the code in `Data#importIntoVectorSchemaRoot` so I just work 
with `ArrowRecordBatch`. 
   
   My use-case is actually Java -> Java, but transferring data over the 
network. It's not a Flight server, but you could imagine if you have Java 
server and Java client, if the Java server gets an `ArrowArrayStream`, if we 
could serialize as-is over the wire, the client would be able to handle that 
and convert to `VectorSchemaRoot` at that point.
   
   Maybe this isn't a valid concern, perhaps the overhead isn't significant 
enough. My assumption was there was a bunch of memory/object allocation 
happening when going to `VectorSchemaRoot` or `ArrowRecordBatch` that was maybe 
unnecessary. I don't know `ArrowArrayStream` looks though compared to the IPC 
stream format.


-- 
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