[
https://issues.apache.org/jira/browse/ARROW-3126?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16594863#comment-16594863
]
Antoine Pitrou commented on ARROW-3126:
---------------------------------------
Ok. The term "read ahead" is a bit misleading, because it implies that I/O is
hidden in the background, which is not how a buffering layer works (the buffer
is filled up synchronously when empty, it's not fed by a separate thread).
Can we-reuse io.BufferedReader for this or is the intention to have a similar
primitive written in C++? Also, does it return a InputStream or a full-blown
RandomAccessFile (the latter is quite a bit more difficult to get right and
optimize).
> [Python] Add buffering option to pyarrow.open_stream to enable larger read
> ahead window for high latency file systems
> ---------------------------------------------------------------------------------------------------------------------
>
> Key: ARROW-3126
> URL: https://issues.apache.org/jira/browse/ARROW-3126
> Project: Apache Arrow
> Issue Type: New Feature
> Components: Python
> Reporter: Wes McKinney
> Priority: Major
> Fix For: 0.12.0
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
