[
https://issues.apache.org/jira/browse/PIG-1168?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Olga Natkovich resolved PIG-1168.
---------------------------------
Resolution: Won't Fix
This is by design. Dump is meant for interactive, not batch mode and as such is
executed right away and not as part of multiquery
> Dump produces wrong results
> ---------------------------
>
> Key: PIG-1168
> URL: https://issues.apache.org/jira/browse/PIG-1168
> Project: Pig
> Issue Type: Bug
> Reporter: Ankur
>
> For a map-only job, dump just re-executes every pig-latin statement from the
> begininng assuming that they would produce same result. the assumption is not
> valid if there are UDFs that are invoked. Consider the following script:-
> raw = LOAD '$input' USING PigStorage() AS (text_string:chararray);
> DUMP raw;
> ccm = FOREACH raw GENERATE MyUDF(text_string);
> DUMP ccm;
> bug = FOREACH ccm GENERATE ccmObj;
> DUMP bug;
> The UDF MyUDF generates a tuple with one of the fields being a randomly
> generated UUID. So even though one would expect relations 'ccm' and 'bug' to
> contain identical data, they are different because of re-execution from the
> begininng. This breaks the application logic.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
