[
https://issues.apache.org/jira/browse/FLINK-12998?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16882779#comment-16882779
]
Alex commented on FLINK-12998:
------------------------------
I'm trying to understand how batch job connectors relate to {{FileSystem}} s
and to Hadoop based ones.
Specifically, I think how to restructure the documentation for the [batch
connectors|https://ci.apache.org/projects/flink/flink-docs-master/dev/batch/connectors.html]:
it mentions file systems, but that information is not updated with newly added
ones.
It also mentions that a required jar should be placed under {{lib}} dir.
Imo, it would be better if file systems information is located in one place
(for example
[here|https://ci.apache.org/projects/flink/flink-docs-master/ops/filesystems/]).
> Document Plugins mechanism
> --------------------------
>
> Key: FLINK-12998
> URL: https://issues.apache.org/jira/browse/FLINK-12998
> Project: Flink
> Issue Type: Sub-task
> Components: FileSystems
> Affects Versions: 1.9.0
> Reporter: Piotr Nowojski
> Assignee: Alex
> Priority: Blocker
> Fix For: 1.9.0
>
>
> Plugins mechanism must be documented before the release. We should write down:
> * benefits
> * how to use existing FileSystem plugins
> * how to implement an own custom FileSystem plugin
> * potential issues of relaying on
> {{Thread.currentThread().getContextClassLoader()}} (currently it's set only
> for {{FileSystemFactory}} class loading and {{FileSystemFactory#create()}}
> method call - if a {{FileSystem}} is accessing {{getContextClassLoader}}
> during runtime (write/reading) it will not work properly as a plugin.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
