AHeise commented on pull request #10482: URL: https://github.com/apache/flink/pull/10482#issuecomment-688348362
There shouldn't be a reason to include our filesystem in the user jar. If you spawn the flink cluster yourself, you should always go with plugin and nothing else. If you don't have the Flink cluster under your control, then the admins must add the plugins (similar to how they added it into lib). Now let's assume that your user code also needs direct access to s3, e.g., to use API directly. Then instead of bundling our jars, use the AWS sdk s3 and bundle it. The whole idea of the plugins is that you would not need to worry about any conflicts. Each plugin / user code can use the library version they want as all are isolated. That wasn't true for jars in `lib`. The only way you interact with the s3 plugin is to use a path starting with `s3://` (or s3a, s3p). ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
