[
https://issues.apache.org/jira/browse/PIG-2659?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13270903#comment-13270903
]
Julien Le Dem commented on PIG-2659:
------------------------------------
Hi Daniel,
Yes my original motivation was to use this in the visualizer.
Note that I made the location transient for now because some tests are
comparing serialized physical plans to a saved binary versions, which is very
brittle and not precise enough. I.E.: it fails whatever you changed to the
physical plan serialized representation.
explanation of the data
M: A[1,4],A[3,4],B[2,4] C: A[3,4],B[2,4] R: A[3,4]
M: Mapper plan contains the following aliases
C: Combiner plan contains the following aliases
R: reduce plane contains the following aliases
each item is:
alias[line,offset]
This could also be split in 3 different properties:
pig.alias.location.mapper: A[1,4],A[3,4],B[2,4]
pig.alias.location.combiner: A[3,4],B[2,4]
pig.alias.location.reducer: A[3,4]
Yes, we should also make sure merged operators info is maintained.
I will take care of the white spaces.
There is a test here:
test/org/apache/pig/newplan/logical/relational/TestLocationInPhysicalPlan.java
> add source location of the aliases in the physical plan
> -------------------------------------------------------
>
> Key: PIG-2659
> URL: https://issues.apache.org/jira/browse/PIG-2659
> Project: Pig
> Issue Type: Improvement
> Components: impl
> Reporter: Julien Le Dem
> Assignee: Julien Le Dem
> Attachments: PIG-2659.patch
>
>
> The goal is to provide better information about what is actually running in a
> job.
> In particular when alias names are being reused.
> For example with the following script:
> {code}
> A = LOAD 'foo' using PigStorage();
> B = GROUP A BY $0;
> A = FOREACH B GENERATE COUNT(A);
> STORE A INTO 'bar';
> {code}
> The job conf will contain the following information
> {code}
> pig.alias.location: M: A[1,4],A[3,4],B[2,4] C: A[3,4],B[2,4] R: A[3,4]
> {code}
> A caveat is that the Logical Plan Optimizer throws away the original
> information when merging Logical Operators.
> this is already the case today with pig.alias
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
