[
https://issues.apache.org/jira/browse/ARROW-5100?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17662122#comment-17662122
]
Rok Mihevc commented on ARROW-5100:
-----------------------------------
This issue has been migrated to [issue
#21587|https://github.com/apache/arrow/issues/21587] on GitHub. Please see the
[migration documentation|https://github.com/apache/arrow/issues/14542] for
further details.
> [JS] Writer swaps byte order if buffers share the same underlying ArrayBuffer
> -----------------------------------------------------------------------------
>
> Key: ARROW-5100
> URL: https://issues.apache.org/jira/browse/ARROW-5100
> Project: Apache Arrow
> Issue Type: Bug
> Components: JavaScript
> Affects Versions: 0.13.0
> Reporter: Paul Taylor
> Assignee: Paul Taylor
> Priority: Major
> Labels: pull-request-available
> Fix For: 0.14.0
>
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> We collapse contiguous Uint8Arrays that share the same underlying ArrayBuffer
> and have overlapping byte ranges. This was done to maintain true zero-copy
> behavior when using certain node core streams that use a buffer pool
> internally, and could write chunks of the same logical Arrow Message at
> out-of-order byte offsets in the pool.
> Unfortunately this can also lead to a bug where, in rare cases, buffers are
> swapped while writing Arrow Messages too. We could have a flag to indicate
> whether we think collapsing out-of-order same-buffer chunks is safe, but I'm
> not sure if we can always know that, so I'd prefer to take it out and incur
> the copy cost.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
