[
https://issues.apache.org/jira/browse/PIG-161?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12590048#action_12590048
]
Pi Song commented on PIG-161:
-----------------------------
I've been thinking about the work distribution layer a bit more. I still have
no idea how this thing could be organized as a layer. In a naive way, we could
compile PO layer to a new graph in a new layer but that would introduce another
unnecessary translation plus some more dummy POs in the current PO layer.
One thing I realized is that only a few operators exhibit properties of work
distribution constructs - LOAD (load + parallelizer), STORE ( stream Aggregator
+ store), and Global Rearranger ( stream aggregator + actual global rearranger
logic + parallelizer). These are only operators to be implemented specific to
the backend. So, why don't we just have these operators as backend specific
operators.
{noformat}
+PhysicalOperator
+Neutral
+HadoopSpecific
+LocalEngineSpecific
{noformat}
@Alan: I will help you think about EvalFunc as soon as possible.
> Rework physical plan
> --------------------
>
> Key: PIG-161
> URL: https://issues.apache.org/jira/browse/PIG-161
> Project: Pig
> Issue Type: Sub-task
> Reporter: Alan Gates
> Assignee: Alan Gates
> Attachments: arithmeticOperators.patch, incr2.patch, incr3.patch,
> incr4.patch, Phy_AbsClass.patch, pogenerate.patch, pogenerate.patch,
> pogenerate.patch
>
>
> This bug tracks work to rework all of the physical operators as described in
> http://wiki.apache.org/pig/PigTypesFunctionalSpec
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
