[
https://issues.apache.org/jira/browse/PARQUET-2222?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17746755#comment-17746755
]
Gang Wu edited comment on PARQUET-2222 at 7/25/23 4:27 AM:
-----------------------------------------------------------
I just revisited this issue. Let me summarize the current state in different
implementations:
* parquet-mr:
** write: Always PLAIN for v1 and RLE for v2.
** read: Use the actual encoding to create decoder, so both encodings are
supported for any data page version.
* arrow-rs:
** write: By default PLAIN for v1 and RLE for v2, but can apply RLE for v1 as
well.
** read: Use the actual encoding to create decoder.
* parquet-cpp:
** write: By default PLAIN for v1 and v2 pages, but both page versions can
apply RLE.
** read: Use the actual encoding to create decoder.
was (Author: wgtmac):
I just revisited this issue and the only problem here and summarize the current
state in different implementations:
* parquet-mr:
** write: Always PLAIN for v1 and RLE for v2.
** read: Use the actual encoding to create decoder, so both encodings are
supported for any data page version.
* arrow-rs:
** write: By default PLAIN for v1 and RLE for v2, but can apply RLE for v1 as
well.
** read: Use the actual encoding to create decoder.
* parquet-cpp:
** write: By default PLAIN for v1 and v2 pages, but both page versions can
apply RLE.
** read: Use the actual encoding to create decoder.
> [Format] RLE encoding spec incorrect for v2 data pages
> ------------------------------------------------------
>
> Key: PARQUET-2222
> URL: https://issues.apache.org/jira/browse/PARQUET-2222
> Project: Parquet
> Issue Type: Bug
> Components: parquet-format
> Reporter: Antoine Pitrou
> Assignee: Xuwei Fu
> Priority: Critical
> Fix For: format-2.10.0
>
>
> The spec
> (https://github.com/apache/parquet-format/blob/master/Encodings.md#run-length-encoding--bit-packing-hybrid-rle--3)
> has this:
> {code}
> rle-bit-packed-hybrid: <length> <encoded-data>
> length := length of the <encoded-data> in bytes stored as 4 bytes little
> endian (unsigned int32)
> {code}
> But the length is actually prepended only in v1 data pages, not in v2 data
> pages.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
