[
https://issues.apache.org/jira/browse/ARROW-14025?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17424124#comment-17424124
]
David Li commented on ARROW-14025:
----------------------------------
If we really want to avoid the dynamic type checking we could instead set the
fragment scan options on the ParquetFileFormat instance when created. I think
{{ParquetFragmentScanOptions$create}} and {{ParquetFileFormat$create}} would
need to gain some options, or we could just make lazy-caching the only
possibility by changing {{dataset___ParquetFragmentScanOptions__Make}} in
dataset.cpp. (This would change all scans, not just ones that go through the
exec plan.)
> [R][C++] PreBuffer is not enabled when scanning parquet via exec nodes
> ----------------------------------------------------------------------
>
> Key: ARROW-14025
> URL: https://issues.apache.org/jira/browse/ARROW-14025
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++, R
> Reporter: Weston Pace
> Assignee: Neal Richardson
> Priority: Major
> Fix For: 6.0.0
>
>
> In ExecNode_Scan a ScanOptions object is built up. If we are reading parquet
> we should enable pre-buffering. This is done by creating a
> ParquetFragmentScanOptions object and enabling pre-buffering.
> Alternatively, we could just default pre-buffering to true for asynchronous
> scans of parquet data.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
