[
https://issues.apache.org/jira/browse/HIVE-7391?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14060330#comment-14060330
]
Gunther Hagleitner commented on HIVE-7391:
------------------------------------------
[~xuefuz], this isn't a big change, but I wanted to bring this up: For stuff
like this it's probably better to copy instead of refactor to common base in
order not to tie Tez and Spark too tightly to the same code. Examples.: Either
of us might want to change the internal representation at some point, or allow
further nesting, change the way we handle the naming of graphs, or change the
translation to execution from topo sorted iteration to something else. If we go
with common base class we'll have to tread carefully not to break each others
code. On the other hand, the amount of duplication is fairly minor. Thoughts?
> Refactoring TezWork/TezEdgeProperty for code reuse
> --------------------------------------------------
>
> Key: HIVE-7391
> URL: https://issues.apache.org/jira/browse/HIVE-7391
> Project: Hive
> Issue Type: Task
> Components: Tez
> Affects Versions: 0.13.0, 0.13.1
> Reporter: Xuefu Zhang
> Assignee: Xuefu Zhang
> Attachments: HIVE-7391.patch
>
>
> Extract DagWork/DagEdgeProperty from TezWork/TezEdgeProperty as common code
> to be reused. Pure refactoring.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
