[
https://issues.apache.org/jira/browse/EAGLE-509?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15447189#comment-15447189
]
Edward Zhang commented on EAGLE-509:
------------------------------------
I am still not convinced by myself for how to do this correctly. We should
think through some problems before we try to fix
1) is that possible we have multiple output streams from one application?
2) is that possible for one output stream to have multiple sink type e.g. kafka
and log file?
3) is the binding between output stream and sink type configurable or fixed in
code?
4) how to configure specific settings for one sink type per output stream
> make streaming application's sink type to be java bean
> ------------------------------------------------------
>
> Key: EAGLE-509
> URL: https://issues.apache.org/jira/browse/EAGLE-509
> Project: Eagle
> Issue Type: Improvement
> Affects Versions: v0.5.0
> Reporter: Edward Zhang
> Assignee: Edward Zhang
> Fix For: v0.5.0
>
>
> Application's sink is used by alert engine, i.e. alert engine needs know each
> field type for one sink.
> Sink type now is a Map type, which means developer has to configure each
> field type in application descriptor file, that is boring and error-prone. If
> we make sink type to be a java bean, then framework can automatically
> generate field type to be used for defining alert metadata.
> Also with java bean, we can use annotation to customize more metadata, for
> example use annotation to specify which field is timestamp column
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
