westonpace commented on issue #34067:
URL: https://github.com/apache/arrow/issues/34067#issuecomment-1421002849

   No, there is no good reason for this.  If the input is a table then the 
dataset writer should be able to retain the field metadata.
   
   The `WriteDataset` API is very generic at the moment.  It takes, as input, a 
scanner.  That scanner can add new columns or remove columns.  So we can't just 
grab the input schema.  Today we just grab the schema metadata instead.
   
   We can change the write node to accept a schema, that shouldn't be too 
difficult. Then perhaps we can create an overload for 
`FileSystemDataset::Write` that accepts a table as input instead of a scanner.


-- 
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