You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@commons.apache.org by "Avin E.M (Jira)" <ji...@apache.org> on 2020/10/22 20:03:00 UTC

[jira] [Comment Edited] (IO-510) ReadAheadInputStream

    [ https://issues.apache.org/jira/browse/IO-510?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17219310#comment-17219310 ] 

Avin E.M edited comment on IO-510 at 10/22/20, 8:02 PM:
--------------------------------------------------------

[~ggregory]:

Indeed my first preference was just to use the on in spark, however, I saw that the implementation in spark has a few limitations:
 1. No support for supplying ExecutorService (minor change)

2. No option to start immediately (also minor change)

3. The read ahead is fixed to just 1 buffer which is not configurable (requires changing some core datastructures)

Because of these limitations, would it be okay if I raised a PR with an implementation which supports these?


was (Author: avin):
[~ggregory]: 

Indeed my first preference was just to use the on in spark, however, I saw that the implementation in spark has a few limitations:
1. No support for supplying ExecutorService (minor change)

2. No option to start immediately (also minor change)

3. The read ahead is fixed to just 1 buffer which is not configurable (requires changing some core datastructures)


Because of these limitation, would it be okay if I raised a PR with an implementation which supports these?

> ReadAheadInputStream
> --------------------
>
>                 Key: IO-510
>                 URL: https://issues.apache.org/jira/browse/IO-510
>             Project: Commons IO
>          Issue Type: New Feature
>          Components: Streams/Writers
>            Reporter: David Mollitor
>            Priority: Major
>         Attachments: IO-510.1.patch, IO-510.2.patch
>
>
> Create a "ReadAheadInputStream".  Such a feature would essentially be a BufferedInputStream that spawns a thread for asynchronous reading.  The thread would continuously fill the buffer (if there was space) from the underlying InputStream.  Thread would wait if buffer was full.



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