[
https://issues.apache.org/jira/browse/ARROW-14740?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17446057#comment-17446057
]
Jonathan Keane commented on ARROW-14740:
----------------------------------------
Yeah, this helper would be relatively lightweight. What you mention about the R
helpers is accurate, it's basically:
* handles creating/using an already created connection (unless it's given in
{{con}})
* handles naming (unless it's given)
* auto-cleanup
* do the actual duckdb registration
The auto cleanup would be nice, but is not strictly necessary. It was
(relatively) easy to wire up in R, so I did, though I did need to use a
slightly different route than normal: see
https://github.com/duckdb/duckdb/issues/2237 and
https://github.com/duckdb/duckdb/pull/2408 which might need to also be exposed
in Python to do it
> [Python] duckdb helper functions
> --------------------------------
>
> Key: ARROW-14740
> URL: https://issues.apache.org/jira/browse/ARROW-14740
> Project: Apache Arrow
> Issue Type: Improvement
> Components: Python
> Reporter: Jonathan Keane
> Priority: Major
>
> In the R package, [we have an integration with
> DuckDB|https://github.com/apache/arrow/blob/master/r/R/duckdb.R] that uses
> the C-Data and C-Stream interface, we include a handful of helper functions
> that handle the conversion for end users (including setting up the DuckDB
> connection, registering the arrow data, etc.)
> Should we also have some helper functions in pyarrow?
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
