[
https://issues.apache.org/jira/browse/TEZ-1169?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14016234#comment-14016234
]
Bikas Saha edited comment on TEZ-1169 at 6/3/14 5:42 AM:
---------------------------------------------------------
Those were random thoughts that I made and ended up distracting the
conversation on this jira. Sorry about that. We can discuss that separately.
Coming back to your suggestion above.
bq. This is effectively dynamic routing, similar to edges - rather than a
static plan where numPhysical inputs is 1 (currently), or hard coded to a
specific value - which is what the current patch does.
bq. Let me see if I can rework this to have additional configuration events
generated via the initializer itself
We are on the same page here. Lets evaluate this alternative. Sounds more
flexible than hard coding the value at compile time.
was (Author: bikassaha):
Those were random thoughts that I made and ended up distracting the
conversation on this jira. Sorry about that.
bq. This is effectively dynamic routing, similar to edges - rather than a
static plan where numPhysical inputs is 1 (currently), or hard coded to a
specific value - which is what the current patch does.
bq. Let me see if I can rework this to have additional configuration events
generated via the initializer itself
We are on the same page here. Lets evaluate this alternative. Sounds more
flexible than hard coding the value at compile time.
> Allow numPhysicalInputs to be specified for RootInputs
> ------------------------------------------------------
>
> Key: TEZ-1169
> URL: https://issues.apache.org/jira/browse/TEZ-1169
> Project: Apache Tez
> Issue Type: Improvement
> Reporter: Siddharth Seth
> Assignee: Siddharth Seth
> Attachments: TEZ-1169.1.txt
>
>
--
This message was sent by Atlassian JIRA
(v6.2#6252)
