[
https://issues.apache.org/jira/browse/ARROW-1409?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16225596#comment-16225596
]
ASF GitHub Bot commented on ARROW-1409:
---------------------------------------
wesm commented on issue #1225: ARROW-1409: [Format] Remove page id from Buffer
metadata, increment metadata version number
URL: https://github.com/apache/arrow/pull/1225#issuecomment-340555574
As long as the JAR version and the Python version are the same, it's okay,
but if users have pyarrow 0.8.0 and connect to a Spark cluster running an older
JAR there will be problems. I don't expect this to become a regular occurrence
-- I don't anticipate the need to break the metadata format again anytime soon
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
> [Format] Use for "page" attribute in Buffer in metadata
> -------------------------------------------------------
>
> Key: ARROW-1409
> URL: https://issues.apache.org/jira/browse/ARROW-1409
> Project: Apache Arrow
> Issue Type: Bug
> Components: Format
> Reporter: Wes McKinney
> Assignee: Wes McKinney
> Labels: pull-request-available
> Fix For: 0.8.0
>
>
> This attribute is currently unused in any Arrow implementation. I think the
> original idea is that the "page" might indicate a particular shared memory
> page, so that a record batch could be spread across multiple memory regions.
> The downside of this unused attribute is that Buffer metadata takes 24 bytes
> instead of 16 due to padding.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
