amoeba opened a new issue, #36115:
URL: https://github.com/apache/arrow/issues/36115

   ### Describe the enhancement requested
   
   The AWS C++ SDK allows configuring log level which is useful when debugging 
parts of the Arrow codebase that use S3. Arrow C++ allows direct control of log 
level during S3 initialization while R gives the user no control and PyArrow 
has a somewhat hard to use helper that allows the user control log level:
   
   ```python
   >>> import pyarrow._s3fs # <--- Note the unusual import with the _
   >>> pyarrow._s3fs.initialize_s3(pyarrow._s3fs.S3LogLevel.Debug)
   ```
   
   In addition to the odd import statement above, the user has to make sure 
they call this early, prior to calling other functions that automatically 
initialize S3 as a side-effect. See 
https://github.com/apache/arrow/issues/35575 for discussion. R does not yet 
have a similar helper but in [the issue to add 
one](https://github.com/apache/arrow/issues/35260) and the linked Pull Request, 
it became apparent that a helper function might not be the best way to let 
users control log level.
   
   Instead of having helpers which may be confusing, an environment variable 
makes a lot of sense as a way to control this instead. There's already been 
some interest expressed in https://github.com/apache/arrow/issues/35575 for 
having an environment variable like `ARROW_S3_LOG_LEVEL` which any 
initialization code would adapted to respect.
   
   I think there are a few sub-tasks here:
   
   - [ ] Add support in C++, Python, and R for `ARROW_S3_LOG_LEVEL`
   - [ ] Include docs for `ARROW_S3_LOG_LEVEL` in C++, Python, R
   - [ ] Add `ARROW_S3_LOG_LEVEL` to 
https://arrow.apache.org/docs/cpp/env_vars.html
   
   
   ### Component(s)
   
   C++, Python, R


-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to