[
https://issues.apache.org/jira/browse/HIVE-4123?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13712916#comment-13712916
]
Owen O'Malley commented on HIVE-4123:
-------------------------------------
Comments:
* merge Utils into SerializationUtils.
* use the zigzag encode/decode in the the SerializationUtils.read/writeVslong
* move Utils.nextLong to the test code
* Utils.getTotalBytesRequired should just use long math. (n * numBits + 7) / 8
should work
* Rename IntegerCompressionReader/Writer to RunLengthIntegerReader/WriterV2
* Create an interface IntegerReader that has:
** seek
** next
** skip
* Make RunLengthIntegerReader and RunLengthIntegerReaderV2 implement
IntegerReader
* The TreeReaders should declare the fields as IntegerReader.
* Each of the startStripe should use the encoding to create the right
implementation of IntegerReader.
* We should do the same with an IntegerWriter interface.
* Replace fixedBitSizes with static methods in SerializationUtils:
** static int encodeBitWidth(int n)
** static int decodeBitWidth(int n)
* Finding the percentiles seems expensive, we should look at an alternative
* Why is the delta blob zigzag encoded? The sign should always be positive or
negative for the entire run.
* Maybe we could create an enum in the Writer that is the version to write that
would look like enum OrcVersion { V0_11, V0_12 } and the StreamFactory could
provide the version to the TreeWriters.
> The RLE encoding for ORC can be improved
> ----------------------------------------
>
> Key: HIVE-4123
> URL: https://issues.apache.org/jira/browse/HIVE-4123
> Project: Hive
> Issue Type: New Feature
> Components: File Formats
> Reporter: Owen O'Malley
> Assignee: Prasanth J
> Attachments: HIVE-4123.1.git.patch.txt, HIVE-4123.2.git.patch.txt,
> ORC-Compression-Ratio-Comparison.xlsx
>
>
> The run length encoding of integers can be improved:
> * tighter bit packing
> * allow delta encoding
> * allow longer runs
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
