[
https://issues.apache.org/jira/browse/PIG-1714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12931543#action_12931543
]
Xuefu Zhang commented on PIG-1714:
----------------------------------
Here is the behavior that Pig is taking:
1. If JVM property "mapred.output.compress" is set to "true", then the output
is always compressed (regardless of the output file extension).
2. If the JVM property "mapred.output.compress" is not set or is set to
"false", then whether pig output is compressed depends on the given file
extension: if the extension is .bz or .bz2, then bzip compression will be used.
If the extension is gz, then gzip compression will be used. In all other cases,
no compression will be performed.
3. When JVM property "mapred.output.compress" is set to "true", then another
property, "mapred.output.compress.codec" must also be set. Otherwise, exception
will be thrown.
> Option mapred.output.compress doesn't work in Pig 0.8 but worked in 0.7
> -----------------------------------------------------------------------
>
> Key: PIG-1714
> URL: https://issues.apache.org/jira/browse/PIG-1714
> Project: Pig
> Issue Type: Bug
> Reporter: Xuefu Zhang
> Assignee: Xuefu Zhang
> Fix For: 0.8.0
>
> Attachments: jira-1714-0.patch
>
>
> Command line options -Dmapred.output.compress and
> -Dmapred.output.compression.codec worked in Pig 0.7, which, when set, would
> compress the output, whether or not the output has an extension .gz, .bz, or
> .bz2. This behavior changed in 0.8 in that compression is on only if the
> output has such extensions. In other words, the command line options have no
> effect.
> Pig needs to clarify the right way to enable/disable compression and
> implement it accordingly.
> The behavior change is probably related to PIg-1533.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
