[
https://issues.apache.org/jira/browse/ARROW-17057?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17567952#comment-17567952
]
Duncan commented on ARROW-17057:
--------------------------------
[~kou]
[https://github.com/apache/arrow/pull/13633]
I don't love it, but it's a minimally invasive way to specify a "stock" AWS
retry strategy, with no change to current behaviour. As a C++/Arrow novice, I
am unsure what best practices/codebase conventions I may have missed here; very
much open to feedback.
> [Python] S3FileSystem has no parameter for retry strategy
> ---------------------------------------------------------
>
> Key: ARROW-17057
> URL: https://issues.apache.org/jira/browse/ARROW-17057
> Project: Apache Arrow
> Issue Type: Bug
> Components: Python
> Affects Versions: 6.0.1, 7.0.0, 8.0.0
> Reporter: Duncan
> Priority: Major
> Labels: pull-request-available
> Time Spent: 20m
> Remaining Estimate: 0h
>
> The Python wrapper for S3Fs does not accept a {{retry_strategy}} parameter,
> but the underlying C++ implementation supports it.
>
> Python wrapper's constructor arguments:
> [https://github.com/apache/arrow/blob/master/python/pyarrow/_s3fs.pyx#L181]
>
> C++ base:
> [https://github.com/apache/arrow/blob/master/cpp/src/arrow/filesystem/s3fs.cc#L729]
>
> The result is that Python users of S3Fs always default to the legacy retry
> strategy, which is very limited.
>
> Suggested fix is to allow the Python wrapper to specify a retry strategy to
> be passed through to the wrapped C++ implementation.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
