[OpenTelemetry] Remove vendored EnvironmentVariablesConfigurationProvider#7146
Conversation
adbb557 to
34d71f7
Compare
martincostello
left a comment
There was a problem hiding this comment.
The change seems OK to me from a code perspective, but I defer to the other maintainers for background on why this was done in the first place that might cause issues (e.g. the additional dependencies for opentelemetry-dotnet-instrumentation).
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #7146 +/- ##
=======================================
Coverage 89.98% 89.98%
=======================================
Files 277 277
Lines 14622 14642 +20
=======================================
+ Hits 13157 13176 +19
- Misses 1465 1466 +1
Flags with carried forward coverage won't be shown. Click here to find out more. |
|
Yeah if the driver is dependency reduction, then we may need to instead consider if we want a sync mechanism for the vendored code intead. I know the instrumentation team were doing work to improve dependency issues, so this may now be suitable. |
|
@open-telemetry/dotnet-instrumentation-approvers perspective: I would not block this change only for AutoInstrumentation. |
|
This PR was marked stale due to lack of activity and will be closed in 7 days. Commenting or pushing will instruct the bot to automatically remove the label. This bot runs once per day. |
|
I can update the tests if we still want to consider this change? I guess we need a final position from the @open-telemetry/dotnet-maintainers on taking this dependency within the SDK. |
|
This PR was marked stale due to lack of activity and will be closed in 7 days. Commenting or pushing will instruct the bot to automatically remove the label. This bot runs once per day. |
|
This PR was marked stale due to lack of activity and will be closed in 7 days. Commenting or pushing will instruct the bot to automatically remove the label. This bot runs once per day. |
|
We discussed this in the SIG just now. We'd like a 👍 from @rajkumar-rangaraj, but so far there's been no dissenting opinions on taking this change. |
alanwest
left a comment
There was a problem hiding this comment.
I recognize the CI isn't passing at the moment, but giving my 👍 based on conversation during our meeting today.
We were originally hesitant to take a dependency on Microsoft.Extensions.Configuration.EnvironmentVariable package because it leads us further down the slippery slope of taking on even more Microsoft.Extensions.* dependencies. If we had it to do all over again, we probably would have avoided all dependencies for the core SDK on Microsoft.Extensions.* packages.
We should continue to take a conservative approach with respect to dependencies, but I don't see a big issue with adding this one. I'd like to hear @rajkumar-rangaraj's input.
Thanks @alanwest for adding the context here. I'm fine with the current approach. |
|
This PR was marked stale due to lack of activity and will be closed in 7 days. Commenting or pushing will instruct the bot to automatically remove the label. This bot runs once per day. |
|
I'll update the tests as recommended and rebase ready for a final review and merge |
- Update dependencies to use Microsoft.Extensions.Configuration.EnvironmentVariables. - Add basic E2E test to validate existing ENV VAR behaviour continues to function.
Fixes #7141
Contributes to #6380 as a pre-requisite simplification
Changes
EnvironmentVariablesConfigurationProvider.Microsoft.Extensions.Configuration.EnvironmentVariables.Merge requirement checklist
CHANGELOG.mdfiles updated for non-trivial changesChanges in public API reviewed (if applicable)