Closed beeme1mr closed 1 month ago
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 97.41%. Comparing base (
5c7bd14
) to head (c4c3d3f
).
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Is this affecting only the Python SDK or is it something that should be specified more generally?
Is this affecting only the Python SDK or is it something that should be specified more generally?
Logging isn't covered in the spec. We could consider general guidance but I would be reluctant to go any deeper than that.
Is this affecting only the Python SDK or is it something that should be specified more generally?
After approving, I'm actually considering if we should specify something here. I'm wondering if this is the wrong solution to the problem.
I think we should be logging errors if a flag is not found, and that this change will make debugging harder. I think this is a situation where we'd want log-spam.
This PR
Impacts
https://github.com/open-telemetry/opentelemetry-demo/issues/1628
Notes
The OTel demo was seeing overly verbose log messages when they disabled a flag. That's because a
FLAG_NOT_FOUND
error was raised by the provider, and the SDK was logging all exceptions. I've updated the SDK to no longer log OpenFeatureErrors since providers commonly raise these, which can lead to log spam. It's worth noting that this information can still be accessed via a hook or evaluation details.