[jira] [Commented] (ARROW-12845) [R] [C++] S3 connections for different providers

["Jonathan Keane (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ARROW-12845?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17364440#comment-17364440 ] 

Jonathan Keane commented on ARROW-12845:
----------------------------------------

This _should_ just work, the trick is configuring the call to `S3FileSystem$create()` correctly. Neal reminded me about https://www.jaredlander.com/2021/03/analyzing-room-temperature-data/#getting-the-data which uses the following setup (you'll need to sub out these values for your values, of course):

{code}
space <- arrow::S3FileSystem$create(
  access_key=Sys.getenv('AWS_ACCESS_KEY_ID'),
  secret_key=Sys.getenv('AWS_SECRET_ACCESS_KEY'),
  scheme="https",
  endpoint_override=glue::glue("{Sys.getenv('DO_REGION')}.{Sys.getenv('DO_BASE')}")
)
space
{code}

Does this not work for you? Do you have the code that you're using that isn't working?


> [R] [C++] S3 connections for different providers
> ------------------------------------------------
>
>                 Key: ARROW-12845
>                 URL: https://issues.apache.org/jira/browse/ARROW-12845
>             Project: Apache Arrow
>          Issue Type: New Feature
>          Components: C++, R
>    Affects Versions: 4.0.0
>            Reporter: Mauricio 'Pachá' Vargas Sepúlveda
>            Assignee: Mauricio 'Pachá' Vargas Sepúlveda
>            Priority: Minor
>
> Hi
> As a part of my thesis, I want to create an S3 bucket on DigitalOcean (what PUC uses), and while I can write parquet files on my laptop and upload to DigitalOcean Spaces (i.e. an "S3 + Google Drive") from the browser or by using rclone, I could work in editing the existing code that allows to connects to Amazon S3, and  provide a function that connects to DigitalOcean/Linode/IBM/etc.
> This could be done in a way that amazon URL is the default and the user could specify something like `new_s3_fun(...,  provider = "Tencent")` and connect to an S3 that is not Amazon.
> Also, this involves the need to write more S3 documentation.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)