[
https://issues.apache.org/jira/browse/ARROW-8486?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17087587#comment-17087587
]
Kazuaki Ishizaki commented on ARROW-8486:
-----------------------------------------
> I don't think parquet::FixedLenByteArray and parquet::ByteArray ever get
> passed to RleDecoder. I'm not sure about Int96.
In my understanding, unfortunately, it occurs at
https://github.com/apache/arrow/blob/master/cpp/src/parquet/encoding.cc#L2530-L2559.
They read to call the constructor of `RleDecoder` at
https://github.com/apache/arrow/blob/master/cpp/src/parquet/encoding.cc#L1484.
Then, `T` is passed to `NextCounts()` thru `GetBatchWithDictSpaced()` at
https://github.com/apache/arrow/blob/master/cpp/src/parquet/encoding.cc#L1502
While Int96 is depreciated, test cases are there. Of course, I can make it
low-priority.
> [C++] arrow-utility-test causes failures on a big-endian platform
> -----------------------------------------------------------------
>
> Key: ARROW-8486
> URL: https://issues.apache.org/jira/browse/ARROW-8486
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++
> Reporter: Kazuaki Ishizaki
> Priority: Minor
>
> The current master causes the following failures in arrow-util-test.
> {code}
> Start 17: arrow-utility-test
> ...
> 17: [ PASSED ] 253 tests.
> 17: [ FAILED ] 11 tests, listed below:
> 17: [ FAILED ] Bitmap.ShiftingWordsOptimization
> 17: [ FAILED ] Bitmap.VisitWordsAnd
> 17: [ FAILED ] BitArray.TestBool
> 17: [ FAILED ] BitArray.TestValues
> 17: [ FAILED ] Rle.SpecificSequences
> 17: [ FAILED ] Rle.TestValues
> 17: [ FAILED ] BitRle.Flush
> 17: [ FAILED ] BitRle.Random
> 17: [ FAILED ] BitRle.RepeatedPattern
> 17: [ FAILED ] BitRle.Overflow
> 17: [ FAILED ] RleDecoder.GetBatchSpaced
> {code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
