Hi Li, I see your point. the reason I put the changes in the `WriteChannel` is 
because that is the single point where the `write` is called on the 
`WriteableByteChannel`. Any channel specific semantics like non-blocking vs 
blocking can be dealt within the `WriteChannel` class - and this class can give 
a solid guarantee to the caller that the data has been written out.  Otherwise 
anyone/everyone who calls write need to loop around the return value. 

[ Full content available at: https://github.com/apache/arrow/pull/2594 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to