[jira] [Commented] (ARROW-1013) [C++] Add asynchronous RecordBatchStreamWriter

2019-09-18 Thread Antoine Pitrou (Jira) 


[ 
https://issues.apache.org/jira/browse/ARROW-1013?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16932297#comment-16932297
 ] 

Antoine Pitrou commented on ARROW-1013:
---

Ah, right. I suppose so. This should be designed depending on the intended use 
cases.

> [C++] Add asynchronous RecordBatchStreamWriter
> --
>
> Key: ARROW-1013
> URL: https://issues.apache.org/jira/browse/ARROW-1013
> Project: Apache Arrow
>  Issue Type: New Feature
>  Components: C++
>Reporter: Wes McKinney
>Priority: Major
>
> We may want to provide an option to limit the queuing depth. The async writer 
> can be initialized from a synchronous writer



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

[jira] [Commented] (ARROW-1013) [C++] Add asynchronous RecordBatchStreamWriter

2019-09-17 Thread Wes McKinney (Jira) 


[ 
https://issues.apache.org/jira/browse/ARROW-1013?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16931679#comment-16931679
 ] 

Wes McKinney commented on ARROW-1013:
-

I'm thinking about an asynchronous IPC writer, not making 
{{OutputStream::Write}} asynchronous. Writing into 
{{RecordBatchStreamWriter::WriteRecordBatch}} does have a non-trivial cost -- I 
guess that we measure it in different scenarios. Ideal context would be Getting 
from one Flight client and Putting into another. Does that make sense?

> [C++] Add asynchronous RecordBatchStreamWriter
> --
>
> Key: ARROW-1013
> URL: https://issues.apache.org/jira/browse/ARROW-1013
> Project: Apache Arrow
>  Issue Type: New Feature
>  Components: C++
>Reporter: Wes McKinney
>Priority: Major
>
> We may want to provide an option to limit the queuing depth. The async writer 
> can be initialized from a synchronous writer



--
This message was sent by Atlassian Jira
(v8.3.2#803003)